1The Knowledge Structure inAmarakoaDoctor of PhilosophyinSanskrit StudiesAuthor:Sivaja S. NairSupervisor:Dr. Amba P. KulkarniDepartment of Sanskrit StudiesSchool of Humanities, University of HyderabadHyderabad 5000462011Contents1 Overview 52 Introduction 82.1 Amarakoa . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.2 Amarasimha - the author of Amarakoa . . . . . . . . . . . . 92.3 Textual Organization . . . . . . . . . . . . . . . . . . . . . . . 122.3.1 Statistics . . . . . . . . . . . . . . . . . . . . . . . . . 122.3.2 Words in Amarakoa . . . . . . . . . . . . . . . . . . . 152.3.3 The First Ka . . . . . . . . . . . . . . . . . . . . . 162.3.4 The Second Ka . . . . . . . . . . . . . . . . . . . 172.3.5 The Third Ka . . . . . . . . . . . . . . . . . . . . 192.4 Commentaries . . . . . . . . . . . . . . . . . . . . . . . . . . . 202.4.1 Amarakoa Translations in Foreign Languages . . . . . 202.4.2 Sanskrit Commentaries on Amarakoa . . . . . . . . . 212.4.3 Other Indian Language Commentaries . . . . . . . . . 253 Lexicon 273.1 The Lexicon . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273.2 Sanskrit Lexicons (Koas) . . . . . . . . . . . . . . . . . . . . 283.3 Modern Sanskrit Lexicons . . . . . . . . . . . . . . . . . . . . 293.4 Electronic Lexicons . . . . . . . . . . . . . . . . . . . . . . . . 293.4.1 WordNet . . . . . . . . . . . . . . . . . . . . . . . . . 303.4.2 IndoWordNet . . . . . . . . . . . . . . . . . . . . . . . 343.4.3 Euro-WordNet . . . . . . . . . . . . . . . . . . . . . . 343.4.4 MultiWordNet . . . . . . . . . . . . . . . . . . . . . . 353.4.5 VerbNet . . . . . . . . . . . . . . . . . . . . . . . . . . 353.4.6 ConceptNet . . . . . . . . . . . . . . . . . . . . . . . . 3514 Gender Information in Amarakoa 374.1 Liganirdhraa in Amarakoa . . . . . . . . . . . . . . . . . 374.1.1 Rpabedhena . . . . . . . . . . . . . . . . . . . . . . . 384.1.2 Shacaryt . . . . . . . . . . . . . . . . . . . . . . . . 394.1.3 Viesavidhi . . . . . . . . . . . . . . . . . . . . . . . 394.1.4 Bhedkhynya na dvandvo . . . . . . . . . . . . . . . 394.1.5 Ekaesa na . . . . . . . . . . . . . . . . . . . . . . . . 404.1.6 Sakara na . . . . . . . . . . . . . . . . . . . . . . . 404.1.7 Gender Indicators in Amarakoa . . . . . . . . . . . . 414.1.8 Tu anta, atha di . . . . . . . . . . . . . . . . . . . . . 414.2 Ligdisagrahavarga . . . . . . . . . . . . . . . . . . . . . . 434.3 Comparison with Monier Williams Dictionary . . . . . . . . . 444.3.1 Monier Williams Sanskrit - English Dictionary . . . . . 444.3.2 The Result of The Comparison . . . . . . . . . . . . . 444.4 Polysemy Distribution in Amarakoa . . . . . . . . . . . . . . 505 Knowledge Structure in Amarakoa 525.1 Amarakoa . . . . . . . . . . . . . . . . . . . . . . . . . . . . 535.2 Organisation of synsets within a varga . . . . . . . . . . . . . 535.2.1 Example 1: Visuh . . . . . . . . . . . . . . . . . . . . 545.2.2 Example 2: Samaya . . . . . . . . . . . . . . . . . . . 545.2.3 Example 3: Ksatriya . . . . . . . . . . . . . . . . . . 565.3 Implicit relations . . . . . . . . . . . . . . . . . . . . . . . . . 575.3.1 Is a part of (avayavvayav) . . . . . . . . . . . . . . . 575.3.2 Is a kind of (parparjti) . . . . . . . . . . . . . . . . 585.3.3 Child-parent relation (janyajanakabhva) . . . . . . . . 585.3.4 Husband-wife relation (patipatnbhva) . . . . . . . . . 595.3.5 Master-possession relation (svasvmibhva) . . . . . . 595.3.6 Livelihood (jvik) . . . . . . . . . . . . . . . . . . . . 616 Ontological Representation scheme for Amarakoa 626.1 Ontology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 636.1.1 Denition of Ontology . . . . . . . . . . . . . . . . . . 636.2 History of Ontology . . . . . . . . . . . . . . . . . . . . . . . . 656.2.1 Western Ontology . . . . . . . . . . . . . . . . . . . . 656.2.2 Indian Ontology . . . . . . . . . . . . . . . . . . . . . 676.2.3 Upper Level Ontologies . . . . . . . . . . . . . . . . . 716.3 Vaiesika Ontology . . . . . . . . . . . . . . . . . . . . . . . . 7226.3.1 Jtibdhaks or impediments for deciding universals . 726.4 Ontological representation scheme for Amarakoa . . . . . . . 756.4.1 Jti and it's classication . . . . . . . . . . . . . . . . 756.4.2 Updhi and it's classication . . . . . . . . . . . . . . 847 Amarakoajnajlam 927.1 Structured Lexicon . . . . . . . . . . . . . . . . . . . . . . . . 927.1.1 Stem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 937.1.2 Amarakoa index . . . . . . . . . . . . . . . . . . . . . 937.1.3 Ligam (gender) . . . . . . . . . . . . . . . . . . . . . 937.1.4 Varga . . . . . . . . . . . . . . . . . . . . . . . . . . 957.1.5 Head Word . . . . . . . . . . . . . . . . . . . . . . . . 957.2 Marking of Various Relations . . . . . . . . . . . . . . . . . . 957.3 Quantitative analysis . . . . . . . . . . . . . . . . . . . . . . . 1007.4 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . 1007.5 Ontological Representation Table . . . . . . . . . . . . . . . . 1017.5.1 Ontological Representation Scheme . . . . . . . . . . . 1017.6 Web application . . . . . . . . . . . . . . . . . . . . . . . . . . 1027.6.1 Web Server : Apache . . . . . . . . . . . . . . . . . . . 1047.6.2 Programming language . . . . . . . . . . . . . . . . . . 1047.6.3 Server side scripting : CGI . . . . . . . . . . . . . . . . 1058 Amarakoa and Hindi WordNet : Synset Comparisons 1068.1 Comparison between Hindi WordNet and Amarakoa . . . . . 1078.1.1 Hindi WordNet Database . . . . . . . . . . . . . . . . 1088.1.2 Amarakoa synset . . . . . . . . . . . . . . . . . . . . 1098.1.3 Comparison and result . . . . . . . . . . . . . . . . . . 1098.2 Semantic changes . . . . . . . . . . . . . . . . . . . . . . . . . 1108.3 Report on various mismatches . . . . . . . . . . . . . . . . . . 1128.3.1 Conceptual Problems . . . . . . . . . . . . . . . . . . . 1128.3.2 Extended usage of the concept . . . . . . . . . . . . . 1138.3.3 Shrink usage of the concept . . . . . . . . . . . . . . . 1148.3.4 Economy and expansion . . . . . . . . . . . . . . . . . 1148.3.5 Diachronic changes . . . . . . . . . . . . . . . . . . . . 1168.4 Suggestions for HWN . . . . . . . . . . . . . . . . . . . . . . . 1178.4.1 Repetition of Synsets . . . . . . . . . . . . . . . . . . . 1178.4.2 Wrong synset . . . . . . . . . . . . . . . . . . . . . . . 1198.4.3 Dierent Concepts but synset members are same . . . 11938.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1199 Conclusion 1219.1 Amarakoajnajla as a model for other koas . . . . . . . . 1219.2 Future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122A WX-notation 123B Amarakoa and MW comparison result 124B.1 Gender dened by words . . . . . . . . . . . . . . . . . . . . . 124B.2 Gender information from word forms . . . . . . . . . . . . . . 125B.3 Compound words . . . . . . . . . . . . . . . . . . . . . . . . . 126C Sample Outputs 127D Amarakoa and Hindi WordNet comparison result 1304Chapter 1OverviewAmarakoa is the most celebrated and authoritative ancient thesaurus ofSanskrit. It is one of the books which an Indian child learning throughIndian traditional educational system memorizes as early as his rst year offormal learning. Though it appears as a linear list of words, close inspectionof it shows a rich organisation of words expressing various relations a wordbears with other words. Thus when a child studies Amarakoa further, thelinear list of words unfolds into a knowledge web. Here we describe oureort to make the implicit knowledge in Amarakoa explicit. A model forstoring such structure is discussed and a web tool is described that answersthe queries by reconstructing the links among words from the structuredtables dynamically.The second chapter provides a brief introduction about Amarakoa, itsauthor Amarasimha, the textual organization of Amarakoa, and thestatistics of each ka. A brief summary of important commentaries onAmarakoa in Indian as well as foreign languages is provided.Chapter three gives a brief introduction about lexicon. Three kinds oflexicons viz. Sanskrit koas, modern Sanskrit lexicons and electroniclexicons are briefed here. The purpose of this chapter is to know the moderntrends in e-lexicon building. An important e-lexicon 'WordNet' is discussedin order to understand various kinds of relations it uses to connect the words.Amarakoa, also known as nmalignusana is primarily considered to bean authoritative tool for knowing the gender of a word and its synonymous5words. In the fourth chapter we discuss the techniques used by Amarasimhafor indicating the gender and number. Amarasimha provides some metarules in the introductory part of the Amarakoa from the third to fth lokas.These techniques are discussed with examples. The other meaning makingkeys viz. tu and atha are also noted. While assigning the genders to variouswords in Amarakoa following Amarakoas meta language, we also noticedsome deviations from the Monier William's dictionary. We carried outthe gender comparison with Monier William's Dictionary and Amarakoa.Finally, the primary purpose of Amarakoa being to provide synonymouswords, we also studied the "polysemy distribution in Amarakoa".The next chapter is named as "Knowledge Structure in Amarakoa". In thischapter we show, through examples, how the organisation of synsets withina varga leads to a semantic web relating various concepts. We take threeexamples from three dierent vargas, representing totally dierent concepts,to show various kinds of relations involved in the consecutive synsets viz.Visu, Samaya and Ksatriya. The detailed study also shows that somesemantic relations among the consecutive synsets are more frequent. Thefrequent implicit relations are : is a part of (avayavvayav), is a kindof (parparjti), child-parent relation (janyajanakabhva), husband-wiferelation (patipatnbhva), master-possession relation (svasvmibhva) andlivelihood (jvik).The sixth chapter, "Ontological Representation scheme for Amarakoa"deals with the ontological representation of each words in Amarakoa.The dierent divisions of Ontology such as Western ontology, Indianontology and upper level ontologies are briefed. We base our ontologicalclassication on Vaiesika ontology. Wherever necessary we also deviate alittle from the Vaiesiaka ontology incorporating classication from otherIndian philosophical schools. In addition to the ontological classication(jti) which shows single inheritance, to capture the imposed or acquiredproperties by various objects, we resort to updhis as well in order to markother properties associated with the object. Jti follows the conditionsimposed by Jtibdhaks while updhi does not.We name the web application showing these relations and ontologicalrepresentation as Amarakoajnajlam. The structure of this systemis discussed in the seventh chapter. Two major parts of this system are6structured lexicon and tables marking various relations. The structuredlexicon have ve parts viz. Stem, Amarakoa index, Ligam (gender),Varga and Head Word. We then discuss the creations of tables markingvarious relations. This is followed by the representation of ontologicalrelations in the data. Finally we brief on the web application with a choiceof 'Apache' server and 'PERL' as a scripting language for CGI.The other task which we carried out is the synset comparisons betweenAmarakoa and Hindi WordNet in order to know how much can weborrow from the existing Hindi WordNet into Sanskrit WordNet based onAmarakoa words there by avoiding duplication of eorts. It is discussed inthe eighth chapter. We rst give the structure of Hindi WordNet database,and the structure of Amarakoa synset. Dierent kinds of the mismatchesand the causes of the dierence in the unmatched synsets are discussedhere. Dierent kinds of problems viz. conceptual problems, extended usage,shrink usage etc. are discussed here and also given some suggestions forHindi WordNet.Some pointers to how Amarakoajnajla can be used as a model for otherkoas and how it can be used for a variety of Natural Language Processingtasks including information retrieval, semantic tagging, disambiguation,ontologies etc. are provided in the concluding chapter.Our main contribution may be summarized as :- to make the implicit relations among various synonymous words inAmarakoa explicit, to carry out this work independently without getting inuenced bythe existing electronic lexicons,and nally to provide a computational model to build similar lexica using otherSanskrit and Indian language koas.7Chapter 2IntroductionStudy of any language involves mastering its vocabulary, grammar andliterature. Since mastering the vocabulary is a hard task, in mother tongueour vocabulary is enriched by practice. But for learning other languagessome means of remembering is required, either by studying the literature ofthat language or versication of words with their meaning. Versication hasbeen successfully accomplished in Sanskrit. Sanskrit scholars quote lexiconssuch as Amarakoa, Vaijayant etc. while commenting on any Sanskrittext as authorities for dierent shades of meaning of such words. TheSanskrit lexicon most widely used by all is Amara's Nmalignusana.Many excellent scholars such as Ksrasvm, Bhnujidksita and so on havecommented upon it.2.1 AmarakoaThe nighau is the oldest lexicographical literature in Sanskrit which ismainly intended to help the interpretation of Vedic texts. The nighausconned themselves to the Vedas which contained not only nominal formsbut also verbal ones whereas the koas dealt with words in classical literatureand restricted themselves to the nominal forms and the indeclinables.The Nmalignusana, the most celebrated and authoritative ancientthesaurus of Sanskrit, authored by Amarasimha is considered as an essentialrequisite for a Sanskrit scholar. It is concise, comprehensive and most8profusely used lexicon in Sanskrit.The Amarakoa (dictionary of Amara) or Nmalignusana is a thesaurusof Sanskrit written by Amarasimha . The word amara means immortaland koa means treasure, casket, pail, collection, dictionary. Thus theword Amarakoa literally means Immortal Treasure. Its alternative namenmalignusana literally means instructions for deciding the genders ofnouns. It is also known as Trika as it contains three kaas. Amarakoais also known as paryyakoa or a dictionary of synonymous words.Amarakoa is prepared most scientically and is comprehensive in allrespects. Sanskrit koas were mainly of two types. Some of the koas dealtwith nominal words while others were developed for explaining the gender.The former is known as Nmamtratantra and the latter is Ligamtratantra.But Amarasimha combined both the styles and methods in his dictionary(Encyclopedia of Indian Literature Vol.2 page 1036). Amarasimha mentionshis work as Nmalignusana in this loka :- 1.1.2 a.k2.2 Amarasimha - the author of AmarakoaAmarasimha , the renowned scholar, is the author of the ancient lexiconAmarakoa, which is clear from the colophon :- This is the only information about Amarasimha that we get from Amarakoa.He didn't mention any information about him anywhere in Amarakoa.Amarasimha was considered as a well-known poet. The following lokareveals his poetic prociency. -9 1It describes the usefulness and derivations of a word. The word 'kavi' in thefourth pda reveals his fame as a poet. The construction of the rst loka ofAmarakoa shows his excellency in poetry2.Mal lintha calls Amarasimha 'kavi' in the last verse of his commentary onAmarakoa. He also says that the immortality of nmalignusana isassured as this faultless work will last as long as the moon and the starsshine in the sky :- 3Amarasimha was a good grammarian. He is counted as one of the eightgrammarians. It was stated in the Kavikahbharaa. 4There is a popular humorous saying that Amarasimha robbed all theauthoritative interpretations of Patajali's Mahbhsya. 5This shows the thorough mastery of Amarasimha in grammar.Amarasimha was the son of abarasvmi and his mother was a dra lady. 61Cited by T.C. Parameshvaran Moosat, 1959, p. 162 (1.1.1 a.k.)3Cited by A.A Ramanathan, 1971, p. XV4Cited by N. P. Unni, 2008, p. IV5Cited by N. P. Unni, 2008, p. V6Indian Catalogue, Vol-6 p. 209, cited Kailash Candra Tripathi, 198410Some scholars conclude that Amarasimha was a Buddhist. Followingarguments help in arriving at this decision:1. In the magalaloka (1.1.1 a.k.)the word jnadaysindhu is very much suitable for Buddha.2. He started the synonyms of devat with the synonyms of Buddha. Innnrthavarga also Buddha is mentioned.If these arguments are accepted, then the following is a counter argumentfor the above.But the word jnadaysindhu can keep as a simile of any God like Visuor iva or Rma or Ksa. If he was a Buddhist then he would havementioned about Bauddhasaghas.Some scholars opined that he was a follower of Jaina as Amarasimhamentioned dharmarjau jinayamau as the synonym of Jaina.Nothing apart from his works is known with certainty about the religionof Amarasimha . He neither mentioned about Buddhasagha's norjainasagha's. It clears that he did not belong to any of these two. Themajor part of his work discusses vartmakavyavasth and yajasamsthand so on. It is very clear from the Amarakoa that he had deep knowledgeof ruti, Smti, Pura, Itihsa, Darana and so on.Some sources indicate that Amarasimha was one of the gem inNavaratnas (nine gems) at the court of the king Vikramditya of 7thcentury, following Klidsa's Jyotirvidbharaam. 7In the Jyotirvidbharaam the courtiers of Vikramditya are :-7Cited by N. P. Unni, 2008 p. IV11

82.3 Textual OrganizationThe Amarakoa consists of verses which can be easily memorized. Most ofthe verses are written in Anusup meter. Even though it is a dictionary ofsynonymous words, a section called nnrthavarga has homonymous wordtoo. The words in nnrthavarga are arranged as per the ending such asknta (words ending with 'ka'), khnta (words ending with 'kha') and so on.Like other Sanskrit texts, Amarakoa also begins with a magalaloka. 1.1.1 a.kBut he doesn't mention any God's name in his magalaloka. Aftermagalaloka he brings up some special rules, which are metarules usefulto understand the gender information of a word. (see the chapter "Genderinformation in Amarakoa").Amarakoa is divided into three kas. Kas are further sub-dividedinto vargas. The rst ka has words pertaining to gods, heavenpacamahbhta (ve basic elements) and abstract concepts such as dik(direction), kla (time), vk, etc. This chapter has ten vargas. The secondka deals with the words denoting real physical objects such as earth,human beings, animals, plants etc. This chapter also has ten vargas. Thethird ka has words related to grammar description of polysemous wordsand other miscellaneous words, and has ve vargas.2.3.1 StatisticsStatistics of Amarakoa is given below. Names of each varga, verse details ineach varga, word number in each varga etc. are described.8Cited by N. P. Unni, 2008 p. IV12KasPrathamaka, dvityaka and ttyaka are the three kas.VargasVargas from each kas are named thus:PrathamakaSvargavarga (heaven)Vyomavarga (sky)Digvarga (direction)Klavarga (time)Dhvarga (cognition)abddivarga (sound)Nyavarga (drama)Ptlabhogivarga (nether world)Narakavarga (hell)Vrivarga (water)DvityakaBhmivarga (earth)Puravarga (towns or Cities)ailavarga (mountains)Vanausadhivarga (forests and medicines)Sihdivarga (lions and other animals)Manusyavarga (mankind)Brahmavarga (priest tribe)Ksatriyavarga (military tribe)Vaiyavarga (business tribe)dravarga (mixed classes)13TtyakaViesyanighnavarga (adjective)Samkravarga (miscellaneous)Nnrthavarga (polysemous)Avyayavarga (indeclinables)Ligdisagrahavarga (gender)lokaslokas in Amarakoa can normally be classied according to their nature,in three classes viz. smnyaloka, niyamaloka and praksiptaloka.Smnyalokas are main verses, which contain synonymous words, and theirmeaning. Niyamalokas describe the meta language and praksiptalokasare the verses which are inserted later by others. To decide whether aloka is praksipta or not, we follow the commentary by Bhnuji Dksita,named Sudhvykhy or Rmram edited by Pait ivadatta in 1915. Ifthe loka is not there in these commentaries, it is considered as praksiptaloka.Verse details of each kas are given below in Table 2.1, Table 2.2 andTable 2.3.Total smnyalokas in Amarakoa are 1,492.1/2Total praksiptalokas are 58.1/2Total niyamalokas are 56.1/2Total lokas in Amarakoa are 1,607.1/214PrathamakaVarga Name Total loka No. Niyamaloka PraksiptalokaSvargavarga 71 5 14Vyomavarga 1.1/2 2Digvarga 35 4Klavarga 31 1.1/2Dhvarga 17 1abddivarga 25.1/2 2.1/2Nyavarga 38 1.1/2Ptlabhogivarga 11 1.1/2Narakavarga 3.1/2 1/2Vrivarga 45 2 1/2Table 2.1: loka statistics of pradhamakaDvityakaVarga Name Total loka No. Niyamaloka PraksiptalokaBhmivarga 18 1 1Puravarga 20 1/2ailavarga 8 1/2Vanausadhivarga 169.1/2Simhdivarga 43 4Manusyavarga 139.1/2Brahmavarga 57.1/2 4.1/2Ksatriyavarga 119.1/2Vaiyavarga 111.1/2 1dravarga 46.1/2 1 1/2Table 2.2: loka statistics of dvityaka2.3.2 Words in AmarakoaAmarakoa contains total 11580 content words from the three kas. Uniquewords from all kas are 9031. Content words are the words that are usedto show the synonymous word or to dene the synonymous words. Uniquewords are the total content words after removing the repetitions in the kas15TtyakaVarga Name Total loka No. Niyamaloka ParaksiptalokaViesyanighnavarga 112.1/2 2Sakravarga 42.1/2 1/2Nnrthavarga 257 1 15Avyayavarga 23Ligdisagrahavarga 46 46 1/2Table 2.3: loka statistics of ttyakaKa Total words Content wordsPrathamaka 2,465 2,300Dvityaka 5,827 5,282Ttyaka 3,288 2,271Total 11,580 9,853Table 2.4: Word statistics of Amarakoaor across the kas. Word statistics according to each kas is given inTable 2.4.2.3.3 The First Ka 1.10.44 a.k.The rst ka contains svargavarga (heaven), vyomavarga (sky),digvarga (direction), klavarga (time), dhvarga (cognition),abddivarga (sound), nyavarga (drama), ptlabhogivarga (netherworld), narakavarga (hell), and vrivarga (water).Varga contents SvargavargaHeaven, Gods, Demons, their arms, ornaments, symbols or vehicles,and other attributes, Fire, Air, Velocity, Eternity, etc. VyomavargaSky16 DigvargaDirections, Deities of the directions, elephants at the points, theirfemale elephants, Cloud, thunder, lightning, rainbow, Rain, hail, rainyday, cloudy day, Moon, types of light, frost, Stars, Planets, sunset,dawn, sunlight, etc. KlavargaTime, day, night, variations of the moon, eclipse, second, hour, months,year, Weather, seasons, Happy, Sorrow, Soul, Mind, etc. DhvargaIndividuality, consciousness, knowledge, sense, organs, tastes,fragrance, colours, etc. abddivargaSarasvat, voice, word, Vedas, Vedgas, stories, legends, sound, typesof sounds, speech, musical sounds, song, ornament's sound, etc. NyavargaSeven musical tones, Musical Instruments, dance, theatrical characters,sentiments, desire, aection, kindness, Festival etc. PtlabhogivargaInfernal region, hole, darkness, Snakes, kinds of serpent, parts of snake,etc. NarakavargaHell, various hells, departed souls, pain, etc. VrivargaWater, Ocean, wave, whirlpool, shore, channel, island, boat, voyage,pilot, deep, sh, sherman, net, sh basket, hook, etc., types of shes,Aquatic animals, crab, turtle, etc. Well, pond, types of ponds, River,Names of rivers, water plants, lotus, water lilly, etc. Parts of theseplants etc.2.3.4 The Second Ka (2.1.1 a.k)17It is divided into ten Vargas or parts. They are bhmivarga (earth),puravarga (towns or cities), ailavarga (mountains), vanausadhivarga(forests and medicines), simhdivarga (lions and other animals),manusyavarga (mankind), brahmavarga (priest tribe), ksatriyavarga(military tribe), vaiyavarga (business tribe) and dravarga (mixed class).Varga contents BhmivargaEarth, land, soil, clay, world, India, regions, types of lands, country,village, kingdom, hill, road, PuravargaCity, suburb, Market, fort, wall, house, kinds of houses, parts of house,house holdings, building land, etc. ailavargaMountains, kinds of mountains, parts of mountains, caves, etc. VanausadhivargaForest, garden, tree, parts of tree, owers, fruits, leaf, shrub, creeper,names of trees, names of shrubs, names of creepers, names of grass, etc. SimhdivargaAnimals, lion, tiger, wolf, deer, kinds of deers, etc. Insects, bee, cricket,birds, hawk, skylark, crow, parrot, etc. parts of birds, wing, beak, etc. ManusyavargaMan, woman, descriptions of woman, blood-relations like son,daughter, husband, wife etc., manhood, dierent stages of manhood,parts of our body, diseases such as cough, scab, etc., dress, ornaments,cloths, types of cloths, cosmetics, fragrant plants, sandal etc., hairdecoration styles, daily usable things etc. BrahmavargaTribes, religious states, sacerdotal, scholars, characters and descriptionsof priests, their occupations and observances, types of res, sacrice, itsrequisites, alms, worship, austerity, study, hypocrisy, marriage, human,pursuits and objects etc.18 KsatriyavargaMilitary tribe, kings, ministers, ocers, servants, enemies, allies,requisites of government, means of defence, and of success, revenue,foresight, insignia of royalty, army, elephants, parts and kind ofelephants, horses, types of horses, vehicles, chariots, litters, warriors,arms and weapons, bow, arrow, javelin etc. war, slaughter, funeral,prison, life, etc. VaiyavargaThird tribe, professions, husbandman, eld, implements of husbandry,corn, pulse, oil-seeds, granary, kitchen, vessels, prepared food, dairy,cattle, trac, weights and measures, commodities, etc. dravargaFourth tribe, mixed classes, artisans, jugglers, dancers, musician,hunters, servants, barbarians, dogs, hogs, theft, nets, fops, loom,plot for burden, wrought leather, tools, art, images, wages, spirituousgaming, etc.2.3.5 The Third Ka (3.1.1 a.k.)Viesyanighnavarga (adjective), sakravarga (miscellaneous),nnrthavarga (polysemous), avyayavarga (indeclinables),ligdisagrahavarga (gender). The third ka contains adjectives,verbs, words related to prayer and business etc..Varga contents ViesyanighnavargaEpithets of persons, Qualities of things, etc. SakravargaMiscellaneous NnrthavargaHomonymous and polysemous words19 AvyayavargaIndeclinables LigdisagrahavargaGenders, Masculine, Feminine, Neuter, Masculine and Feminine,Masculine and Neuter, Feminine and Neuter, Three genders, variationsof gender.2.4 CommentariesAmarasimha 's lexicon is the oldest work of the kind now extant. It is ofgreat interest to note that, it has been universally accepted as an authorityby the Brahmas and the Jainas alike. The fact that it has been commentedupon by Buddhists like Subhticandra, by Jainas like dharapaita andNcirja, and by Brahmans like Ksrasvmin, Mallntha and Appayyadksitatestied to its usefulness to every class of Sanskrit students.The commentaries on Amarakoa are available in almost all Indian languages.Translations of the Amarakoa into Chinese, Tibetan, Italian, French,Mongolia, Burmese etc. have been recently discovered. It is dicult toprovide the exact number of commentaries on Amarakoa, as many of themare not available. In the world of Sanskrit literature a separate section isdedicated to the study of Amarakoa such as commentaries in Sanskrit andother languages, translations in other languages, other koas on the basis ofAmarakoa, etc. M.M Patkar in his book History of Sanskrit Lexicographymentions nearly 60 commentaries9on Amarakoa and also he tells that "Dr.Aufrecht records not less than forty commentaries on it in his CatalogusCatalogorum10.2.4.1 Amarakoa Translations in Foreign LanguagesChinese translation of Amarakoa is written by Guarata in 6thCenturyA.D. The Italian translation of Amarakoa is published in 1798. Frenchtranslation by ALA Loiseleur-Deslongchamps is published at Paris in1839-1845. Tibetan translation of Amarakoa was done by Krticandra and9M.M Patkar, 1981, p.172-17410M.M Patkar, 1981, p.1920Grags-pargyal-mthsan of Yar-lus at Yam-bu, the ancient capital of Nepal11,and it is published in 1912.12Burmese Amarakoa was written (calligraphed) in 1938 by Gulakra atthe Shwegu Hall. Its photo-mechanical reprint is done by Lokesh Chandra,son of the most famous lexicographer Raghu Vra. He mentioned in theintroduction of his book that his father has collected Amarakoa in dierentlanguages of Asia. From them he reproduced the Tibetan rendering of Si-Tuthe well-known grammarian of the seventeenth century13.2.4.2 Sanskrit Commentaries on AmarakoaSeveral well-versed Sanskrit commentaries emerged on Amarakoa. Some ofthese commentaries are commented from some special point of view. E.g.Vykhysudh is commented on grammatical point of view. ksarvasvamgives more elaborated grammatical details than Vykhysudh. Details ofsome famous commentaries on Amarakoa are given below.Rmram or VykhysudhVykhysudh or Rmram is a commentary of Amarakoa written byBhnuji Dksita, son of the celebrated grammarian Bhaoji Dksita. It isalso known as Bhnujibhayam in the name of the author. It is a wellknown and most celebrated commentary of Amarakoa. The magalalokaof Sudhvykhy itself says that Bhaoji Dksita is his teacher and he is goingto write this commentary on the basis of the opinions of Pini, Ktyyanaand Patajali the Munitraya of Sanskrit grammar. The loka is - ( ) () (1 v.s)The colophon of Sudh says that Bhnuji Dksita had written this book withthe desire of Krtisimhadeva, ruler of mahdhara and the prince of Vaghelafamily.11Claus Vogel, 1979, p.31212A.A Ramanathan, 1971, Introduction, p.xvi13Lokesh Chandra, 198421 It is also clear that Bhnuji Dksita was the son of Bhaoji Dksita thewell-known grammarian. Sudh itself shows his grammatical scholarshipas he includes the root of the words, its gaa, pad like tmanepad orparasmaipad etc. And it's i vyavasth like se, ani or ve and the sux,it's Pinya stra, sutra number according to Asdhyy, the vigrahavkyaof the word, etc.. He also makes lexicological references like Medinkoa,Trikaesa, vivakoa etc. Some times Bhnuji Dksita includes the localname of the word also eg. 'jh' iti khyty, 'dopahariy' iti khytasyaetc. We can feel that Bhnuji Dksita's deep grammatical knowledge andthe familiarity with other koa's which were available at that time throughthe study of Sudhvyakhy.For our work, we relied on this work a lot and in case of conict we resortedto this commentary.AmarakoodghanaIt is a commentary on Amarakoa written by Ksrasvmin. Nmaparyaaor amarakoodghaana are the other names of this commentary. It is a veryold commentary which is available now.ksarvasvamIt is also a well-known commentary on Amarakoa, written by VandyaghayaSarvnanda, a Bengali scholar. Following is the beginning loka of thiscommentary. It states that, at the time of Sarvnanda, ten commentaries existed andstudied by the commentators. As the commentator was not satised withany of these ten, he was forced to write a new commentary. The endingverses of the commentary :-22 With these verses he states the eort he had taken to write this commentary.The three grammarians referred by him are Pini, Ktyyana and Patajali.He refers poets like Klidsa, Bhsa, Ba, Mgha, Bhavabhti, Murri,Vikhadatta, Vysa, Vlmki etc. We can feel his knowledge in grammarand literature and critical thoughts of the writer through the study of thiscommentary.AmarapadaprijtaAmarapadaprijta the commentary of Amarakoa written by Mallintha, isthe commentator of pacamahkvya's. He belongs to the Andhra country,as testied by the usages of Telugu equivalents in his commentary. His Fatherwas Nsimhasrin, and he is from the family of Bollinmi. ..........14Mallintha gives Telugu equivalent to almost all words in his commentary.He uses lots of citations also. Some authors mentioned that some of thecitations are not traceable in the Catalogues Catalogorum also. He referrednearly 200 books in his work. The highest citation in the rst two kasare from Vaijayantikoa of Ydavapraka. He gives nearly 213 citationsfrom this book. Mallintha's knowledge of grammar is very deep, and alsoit enables him to point out some points which were missed by many othercommentators. The commentary Amarapadaprijta of Mallintha is veryrich in case of word knowledge.AmarapadavivtiAmarapadavivti is the commentary written by Ligayyasrin. He is alsoknown as Liga or Ligabhaa and his work is known as Ligabhayam.Like Mallintha he also belonged to Andhra Pradesh. According to thecolophons of Amarapadavivti his father's name is Kmaya Bhaa.14A. A. Ramanathan, 1971, p. xlvi23Ligayyasrin speaks respectfully about Ksrasvmin in the introductoryverses of his commentary. He carefully memorized Amarakoodghanambefore writing his commentary. The verse is - Ligayyasrin explains the derivations and meanings of amara-words. Healso gives the Telugu equivalent words in his commentary. His commentarywas well received all along, especially in the southern region of India.AmarapadavivekaAmarapadaviveka is written by Mahevara. It is also well-known in the nameof the commentator Mahevara as Mahevaratk. He was the native ofMaharashtra as is obvious from his use of Marathi words to explain theAmara words. He didn't mention any information about him anywhere inhis book. Amarapadaviveka is a good and famous commentary on Amarakoa.PadacandrikPadacandrik is authored by Ryamukua whose surname is Bhaspati.He belongs to Bangala. He mentioned around 16 earlier commentatorsin his work that were available at his time and he repeatedly referredto some of these commentaries. According to H.T Colebrooke15thosecommentators are :- Ksrasvmin, Subhti, Ha Candra, Kaliga, Kokaa,Sarvadhara, Govardhana, Drvia, Bhojarja, Rjadeva. (ksarvasva,Vykhymta, Mdhav, Madhumdhav, Abhinanda, Sarvnanda - These arecommentaries). Padacandrik is one of the full commentary on Amarakoa.These are some other well-known commentaries on Amarakoa.Budhamanoram by Vedntimahdeva, Amarapysa by Rmaksadksita,Amaracandrik by Paramnandamaidhila, Amarapadamukura byRagcrya, Amarakoakaumud by Nryaa arman, Amarapacikby Nryaa, Kmadhenu by Subhticandra, Kriykalpa by dhara,15Kosha or Dictionary of the sungskrita language by Umura singha with an Englishinterpretations and annotations, Preface, vii.24Gurublaprabodhik by Vemkaevarayajvan, Trikacintmai byRaghundhacakravartin, Trikaviveka by Rmandhavidyvcaspati,Padamajar by Rmevara arman, Padrthakaumud by NryaaCakravarttin, Bhvin by Bhavanadsa, Mugdabodhin by Bhaasena,Ligabhaya by Ligabhaa, Subodhin by Jtnu Dksita2.4.3 Other Indian Language CommentariesAlmost in all Indian languages, the commentaries on Amarakoa areavailable. These are of dierent types, some are the translations of somefamous Sanskrit commentaries, some are Indian language commentariesbased on other Sanskrit commentaries.Malayalam commentariesPramevaryamParamevaran mssat, the writer of the Malayalam commentary ofAmarakoa named Pramevaryam, has written three more Malayalamcommentaries for Amarakoa. He is a good critique of Amarakoa. Hefollowed Bhnujidksita in his commentaries, but some times he alsocriticized Bhnujidksita.Pramevaryam is the most popular and famous commentary of Amarakoain Malayalam. It has rich grammatical properties. For each word heprovides these properties, it's endings, gender, case, number, explanationin Malayalam, another readings (phntara) and also quotes other lexiconsand books as reference.TriveParamevaran mssat's another commentary for Amarakoa known asTrive is also a famous Malayalam commentary. Like Pramevaryam thisis not a deep commentary. In this he explains only the word's meaning inMalayalam and some times in English also. This commentary is very usefulfor a person who is in search of some general information on Amarakoa.Samksiptapramevar and Padrthadpikvykhy are the other Amarakoacommentaries of Paramevaran mssat25Blapriy is also a famous commentary on Amarakoa in Malayalam writtenby Kaikkulagara rma vrier. He himself was a teacher, so he created thiswork for the easy understanding of students.26Chapter 3LexiconThe history of lexicon begins from the vedic period itself. The very rstlexicon is Yska's Nighau, and it is the only lexicon which is available nowfor vedic words. According to the derivation of the word Nighau, given byYska in his Nirukta - the Nighau comprises of a list of vedic words1.Syacrya in his gvedabhsyopakrama, dene Nighau as, "a book wherethe same meaning -synonymous- words are considered as a group". He refersto ten Nighaus2. Now a days the term Nighau is being used in the senseof "dictionary" in many modern Indian languages. Koa or Kosa are thewords used for lexicons in the Sanskrit literature.3.1 The LexiconLexicon is a vocabulary of a language, including its words and expressions.Lexeme is the fundamental unit of the lexicon of a language. Lexeme is aword or stem that is a meaningful unit in a language and coincides with theabstract unit underlying a given set of inected forms. It is an abstract unit1 ( .)2 , , , ( , , ., .)27of morphological analysis in linguistics. A lexical database is an organizeddescription of the lexemes of a language.Lexicon is "the collection of words". Dictionary, Thesaurus and Encyclopediaare dierent types of lexicons. Dictionary - collection of words and its meanings, some times examples,which are arranged in alphabetical order. Thesaurus - collection of words and its synonyms, which are arrangedfollowing some structure, such as ontological classication. Encyclopedia - collection of words which are arranged in alphabeticalorder. Each entry in an encyclopedia consists of an essay describingvarious aspects of the entry.Koa literature in Indian tradition is very rich and at the same time it haslot of variety also. Koas are typically organised following an ontologicalstructure. Since the koas were memorised orally, in order to facilitate thememorisation with ease, most of these follow metrical compositions.3.2 Sanskrit Lexicons (Koas)In the book "Indian Lexicography", Claus Vogel explains the characteristicsof Indian lexica as - "Indian dictionaries may be synonymic or homonymic.The synonymic dictionaries are systematic catalogues of words with one andthe same meaning (ekrtha, samnrtha); A neat and proper discriminationof both categories is not always practicable since many synonymic lexicainclude a homonymic section or chapter as well."3Some of the synonymous koas are Nmaml, abdaratnkara,abdacandrik etc.. Homonymous are Anekrthasamuccaya,Anekrthadvanimacar, Vivapraka etc.. Dhanvantarinighau,Sabdacandrik, Rjanighau come from the domain of medicine.Praspraka of Vedgarya covers the domain of Indian astronomyand astrology. The Rjavyavahrakoa of Raghuntha presents technical3A History of Indian Literature; Vol. 5, Claus Vogel, P.30528terms used in administration.Lexicons which have combination of both synonymous and polysemous wordsare Amarakoa, Vaijayantkoa, Abhidnacintma etc.. Some focus onetymology or some on grammar while some are arranged on the basis ofsyllables, and so on.3.3 Modern Sanskrit LexiconsThe method of representing the lexical units in metrical style graduallyfaded in the modern times, because without the help of any commentaries toget the proper meaning of the word became very dicult. This changed thecontent of the dictionaries and also its organisation. The new representationstyle brought in clarity.H. H. Wilson's A Sanskrit-English Dictionary, Sanskrit-English Dictionaryby Theodore Goldstucker, A Sanskrit-Latin Dictionary by Bopp, DictionaryClassique Sanscrit-Francais of Emile Burnouf etc. are the oldest bilingualdictionaries.Vcaspatyam compiled by Taranatha Tarkavacaspati, abdakalpadruma byRaja Radhakanta Deva etc. are some monolingual Sanskrit dictionaries.Monier Monier-Williams' Sanskrit-English dictionary and English-Sanskritdictionary, The Practical Sanskrit-English Dictionary compiled rst by V.SApte etc. are the famous modern Sanskrit bilingual dictionaries.3.4 Electronic LexiconsThe aim of an electronic lexicon is dierent from that of printed dictionaries.The entries of an electronic lexicon, should contain all orthographical orinectional variations. Since programming languages handle simple one ortwo dimensional data structures more easily, embedded entries as in printeddictionaries should be avoided for an electronic lexicon. Information forgenerating all the derivative forms is necessary, whereas gloss or denition29may not necessarily be a top priority.WordNets are the most popular electronic lexicons. WordNets are availablefor many languages like English, Italian, German, Spanish, Portuguese,Hebrew, Romanian, Latin etc. Indian languages like Hindi, Sanskrit,Marathi, Telugu, Tamil, Malayalam, Konkani etc..VerbNet, ConceptNet, PropNet, FrameNet etc. are the other famouselectronic lexical databases.3.4.1 WordNetWordNet is an electronic lexical database. It is a large lexical database ofEnglish language. This work was inspired by the psycholinguistic theory ofhuman lexical memory. Nouns, verbs, adjectives and adverbs are organisedin sets of synonyms, each of which represents a lexical concept. Thesesets of synonyms are interconnected by a certain number of relations andorganised into taxonomies. WordNet distinguishes two types of relations:lexical relations, such as synonymy, antonymy, polysemy etc., and semanticrelations, such as hypernymy, hyponymy, holonymy, meronymy etc..WordNet was developed at Princeton University's Cognitive ScienceLaboratory. Over the years many linguists, lexicographers, students, andsoftware engineers have contributed to the project. These group started todevelop this lexical database in 1985.Word forms in WordNet are represented in their familiar orthography; wordmeanings are represented by synonym sets (synsets) - lists of synonymousword forms that are interchangeable in some context. Synset is the basicunit of WordNet. It groups English words into sets of synonyms calledsynsets, and provides short, general denitions, with some examples.E.g. Synset :: {example, illustration, instance, representative}Denition :: an item of information that is typical of a class or group;Examples :: "this patient provides a typical example of the syndrome";"there is an example on page 10".30Minimality, coverage and replaceability are the three principles thatdetermine a synset:1. Minimality: Only the minimal set that uniquely identies the conceptis used to create the synset, E.g., {ghar, kamaraa} (room)ghar which is ambiguous is not by itself sucient to denote theconcept of a room. The addition of kamaraa to the synset brings outthis unique sense.2. Coverage: The synset should contain all the words denoting a concept.The words are listed in order of (decreasing) frequency of theiroccurrence in the corpus. {ghar, kamaraa, kaksh} (room)3. Replaceability: The words forming the synset should be mutuallyreplaceable in a specic context. Two synonyms may mutually replaceeach other in a context C, if the substitution of the one for the otherin C does not alter the meaning of the sentence. Consider,Synset : {svadesh, ghar} (motherland) {apanaa desh} (thecountry where one is born)Example : amerikaa meN do saal bitaane ke baad shyaamsvadesh/ghar lauTaaGloss : America in two years stay after Shyam motherlandreturned.English Translation : Shyam returned to his motherland afterspending two years in America4WordNet has information belonging to four Parts of Speech of EnglishLanguage, viz nouns, verbs, adjectives and adverbs. The other Parts ofSpeeches viz prepositions, conjunctions, pronouns, etc. are not covered inWordNet.4Kulkarni et.al, 2009,31Three types of words are there in WordNet. Polysemous, Homonymousand Monosemous. Polysemous words refers to a word with multiple relatedmeanings with same spelling. Homonymous words have the same spellingand the same pronunciation but have dierent meanings. Words with singlemeaning are termed as monosemous.WordNet records various relations between synonym sets or synsets. Twokinds of relations are recognized: lexical and semantic. Lexical relationshold between word forms, it is word - word relation; and semantic relationshold between word meanings or concepts, it is meaning - meaning relation.Semantic relations are the dominating relations in WordNet.Semantic RelationsSemantic relation is a relation between meanings or concepts, that arerepresented by synsets.Hypernymy and hyponymy, meronymy and holonymy, troponymy andentailment, attribute, similar-to, etc. are some semantic relations. Hypernymy and Hyponymy If synset 'A', is a kind of synset 'B' then 'A' is the hyponym of'B' and 'B' is the hypernym of 'A'.* E.g.: chloric acid is hyponym of acid and acid is thehypernym of chloric acid Holonymy and Meronymy Synset 'A' is a meronym of synset 'B', if 'A' is a part of 'B',conversely 'B' is a holonym of 'A', if 'B' has 'A' as a part.Holonyms and Meronyms can be as three pairs:-* Member of and Has MemberE.g.:- Island is a member of archipelago, archipelago hasmember as an island.* Substance of and Has substanceE.g.:- Paper has substance cellulose, cellulose is asubstance of paper.32* Part of and Has partE.g.:- Chapter is a part of text, text has part as chapter. Troponymy and Entailment These are verb relations. Troponymy is a particular kind ofentailment. Activities which are temporally co-extensive arerelated as Troponymy.* E.g.:- limp and walk (limping is the temporally co-extensivework of walking. So it became the troponym of walking)* lisp and talk Activities which are having, proper temporal inclusion, are relatedas Entailment.* E.g.:- snore and sleep (the sense snore have proper temporalinclusion in the sense sleep. So snoring is the entailment ofsleeping)* buy and pay Attribute It links together synset 'A' with an attribute synset 'B', when 'B'is a value of 'A'* E.g.:- measure - is the attribute of standard and nonstandard.(standard and non standard are values of measure) Similar-to It links two adjective synsets.* E.g. (1):- Synset1 is {last}, and meaning is 'Immediately Past'andSynset2 is {Past}, and meaning is 'Earlier than the presenttime'.* E.g. (2):- Synset1 is {last}, and meaning is 'occurring at thetime of death'.andSynset2 is {dying}, and meaning is 'in or associated with theprocess of passing from life or ceasing to be'33Lexical RelationsLexical relations are the relations between word or words in a synset.Antonymy relation is a lexical relation, which relates between two wordsfrom two dierent synsets. Rest of the words from the synset will not haveany eect of the lexical relations.Antonymy is the famous lexical relation in WordNet. Pertainym of, etc. arethe other lexical relations. Antonymy It links together two words that are opposites of each other.* E.g.:- beauty and ugliness* E.g.:- come and go Pertainym of It is an adjective relation. An adjective 'A' is related to anotheradjective or to a noun 'B' if 'A' pertains to 'B'.* E.g.(1):- bicentennial(adj) pertains to centennial(adj) pertainsto century(n)* E.g.(2):- animatedly(adv) pertains to animated(adj)3.4.2 IndoWordNetIndian language WordNets are collectively called as IndoWordNet. It is alsoa multilingual lexical database. 16 out of 22 ocial languages of India,have started making their WordNets under the leadership of IIT Bombay.These languages are: Hindi, Marathi, Konkani, Sanskrit, Nepali, Kashimiri,Assamese, Tamil, Malyalam, Telugu, Kannada, Manipuri Bodo, Bangla,Punjabi and Gujarati. These languages cover the length and breadth ofIndia.3.4.3 Euro-WordNetEuroWordNet is a multilingual lexical database for several Europianlanguages. This is the rst multilingual electronic lexicon. The aim of theEuroWordNet-project is the development of a database with WordNets for34English, Spanish, Dutch and Italian, similar to the Princeton WordNet1.5,which contains basic semantic relations between words in English. TheDutch, Italian and Spanish WordNets will be linked to the WordNet1.5 usingequivalence relations. The resulting multilingual database can directly beused in (multi-lingual) information retrieval.3.4.4 MultiWordNetMultiWordNet is a multilingual lexical database including informationabout English and Italian words. It is an extension of WordNet 1.6, alexical database for English developed at the Princeton University. TheItalian synsets are created in correspondence with the Princeton WordNetsynsets, whenever possible, and semantic relations are imported from thecorresponding English synsets; i.e., we assume that if there are two synsetsin Princeton WordNet and a relation holding between them, the same relationholds between the corresponding synsets in Italian. The Spanish, Portuguese,Hebrew, Romanian and Latin WordNets are the member languages of theMultiWordNet model.3.4.5 VerbNetVerbNet is the largest on-line verb lexicon currently available for English. It isrst released in 2005. It is a hierarchical domain-independent, broad-coverageverb lexicon with mappings to other lexical resources such as WordNet,FrameNet etc.. VerbNet is organized into verb classes extending Levin(1993)5classes through renement and addition of subclasses to achievesyntactic and semantic coherence among members of a class. Each verb classin VerbNet is completely described by thematic roles, selectional restrictionson the arguments, and frames consisting of a syntactic description andsemantic predicates with a temporal function.3.4.6 ConceptNetConceptNet6is a commonsense knowledge base and natural languageprocessing tool-kit which supports many practical textual-reasoning tasks5Verb index from "English Verb Classes And Alternations: A PreliminaryInvestigation", by Beth Levin, published by The University of Chicago Press, 1993.6http://web.media.mit.edu/hugo/conceptnet/35over real-world documents including topic-gisting, analogy-making, and othercontext oriented inferences. The knowledge base is a semantic networkpresently consisting of over 1.6 million assertions of commonsense knowledgeencompassing the spatial, physical, social, temporal, and psychologicalaspects of everyday life. ConceptNet is generated automatically from the700,000 sentences of the Open Mind Common Sense Project - a WorldWide Web based collaboration with over 14,000 authors. Fig.3.1 shows theConceptNet representation of - "An excerpt from ConceptNets semanticnetwork of commonsense knowledge. Compound (as opposed to simple)concepts are represented in semi-structured English by composing a verb(e.g. drink) with a noun phrase (coee) or a prepositional phrase (inmorning)"7.Figure 3.1: Knowledge representation in ConceptNet7Liu,H, 200436Chapter 4Gender Information inAmarakoaAmarakoa is primarily named as Nmalignusana. The meaningof the name is "a work that deals with instructions related to thegender of nouns". This book in agreeing with the true meaning of thename, Nmalignusana, gives the gender instruction of nouns. In theintroduction part of the book, Amarasimha gives the general guidelines.Through these verses, the normal rules for deciding the gender of a nounis explained. Through out Amarakoa some gender indications are givenwherever the explanations are needed. These verses and the gender indicatingwords within Amarakoa from a meta-language. In what follows we describethis meta-language.4.1 Liganirdhraa in AmarakoaAmarasimha lists nominative singular (pradhamaikavacana) words throughthe verses of Amarakoa . For some exceptional cases like (2.6.6 a.k.) he uses the special form like dr and indicatesit's specialty like pubhmni also, because the stem dr will always usein masculine-plural. For illustrating the gender, Amarasimha uses specicwords such as pu, str, astr, napu, etc. which t extremely well in theverse. He describes the rules as :- 37 (1.1.3 a.k.) (1.1.4 a.k.) (1.1.5 a.k.)In these lokas he describes various techniques he employed to code thegender information like rpabedha, shacaryam, viesavidhi etc. Wediscuss each of these with examples.4.1.1 RpabedhenaFrom the word form (1.1.28 a.k.)Here kaumodak and gad are in feminine gender and khaga and nandakaare in masculine gender. He didn't mention any information about thegender of these words, but it should be inferred using our world knowledge.Qualier-Qualicand relation (2.6.90 a.k.)It is very dicult to decide the gender of the word hanu. Because wordsending in 'n' have the same form in masculine and feminine gender innominative singular case. Here the adjective tatpara qualies hanu whichis in masculine gender. There is a rule that qualier and qualicand shouldbe in same case, number and gender. 1From this rule we can conclude that the word hanu is in masculine gender.1Cited by Trntha Tarka Vcaspati, Vol.6, p.491338Pronoun (2.9.33 a.k.)The words pumn and kutupa are masculine. So one may get confused withthe gender of the words kut and ktte. But the pronoun s claries it andremoves the confusion and the gender of these two words is xed as feminine.4.1.2 ShacarytIn case of (1.3.33 a.k.), bhnu can be both masculine as well asfeminine. The proximity of kara indicates the gender of bhnu should bemasculine.4.1.3 ViesavidhiSometimes Amarasimha himself uses some special words to indicate thegender wherever it is not clear from any of the above means. This is the"viesavidhi" in Amarakoa . E.g. (1.7.6 a.k.) etc. Herethe word bher is feminine and dundubhi is masculine.4.1.4 Bhedkhynya na dvandvoAmarasimha frequently uses coordinative compounds in his lokas. But whilemaking compounds he follows some rules. Pini allows in his grammar thatdierent gendered words can be a part of coordinative (dvandva) compound.But in Amarakoa the words which have gender number combinations(feminine plural, feminine dual etc.), or more than one gendered words arenot included in coordinative compound. For example :- (1.1.11 a.k.)Here apsaras have a special gender viz. feminine plural, than other words inthis coordinative compound. Considering only this line it is dicult to decidethe gender of the word apsaras. Here he treated as aco-coordinative compound, but in the following line it is mentioned that thegender of the word apsaras is feminine plural.39 (1.1.52 a.k.)Here the confusion for desiding gender is solved, using viesavidhi.4.1.5 Ekaesa naIn case of the words which are in dierent gender he avoids ekaesa. E.g. (3.3.232 a.k.)Here, the rst nabha is the synonym of sky and it is in neuter gender. Thesecond nabh is the synonym of rvaa and it is in masculine gender. Theverse could be composed as . Then the paraphrase would be and the gender will be neuter only as per the Pinian rule 2. If he uses ekaesa principle, the information may belost. To get nabha in both masculine and neuter Amarasimha does not useekaesa.4.1.6 Sakara naIf synonyms are in dierent genders Amarasimha arranges words in a specicorder. For example :- (1.6.11 a.k.)In this verse stava is in masculine gender, stotra is in neuter gender, andstuti and nuti are in feminine gender. Had he composed as then it will become very much confusing as regards gender identication.Amarasimha didn't follow any rule that a particular gender word should occurrst. E.g. :- consider 1.4.30 a.k.Here janurjananajanmni is in neuter gender, janirutpatti is in femininegender udbhava is in masculine gender.If a word has three genders the word trisu is used. E.g. :- (1.10.7 a.k.)2 2/4/2640taa is in masculine, ta is in feminine, taa is in neuter genders. Theword dvayo indicates the word has masculine and feminine genders. E.g. :- (1.1.47 a.k.)It means the word avini is in both masculine and feminine genders. If theindicator word has negation then it indicates the other two genders. E.g. :- (2.2.6 a.k.)the word uaja is astriym i.e masculine as well as neuter, and not feminine4.1.7 Gender Indicators in Amarakoastr and stry for feminine genderpum and pusi for masculine gendernapu and klbam for neuter genderastr, na str and punnapu for both masculine and neutergendersstrnapu for both feminine and neuter gendersstrpu, dvayo, dve and dvau for both feminine andmasculine genderstrisu for all the three gendersand avyaya for indeclinablen is also used by Amarasimha to indicate masculine gender E.g.:- 1.1.71 a.k.bhmni and bahusu are used for indicating bahuvacana. E.g. :- (2.6.6 a.k.)and (1.1.52 a.k.)These are the gender indicators used by Amarasimha.4.1.8 Tu anta, atha diIn lokas the words tu and atha are to be connected with the consecutiveword for getting the meaning. "tu anta" indicates the token will take placeafter tu, and the preceding word will be the meaning of the token, like :-41nagartvamarvat. Here Amarvat is the token and nagar is the meaning.In the case of atha di, before atha the token will take place and thesucceeding word will be the meaning of the token. E.g. javo'tha ghram.Here java is the token and ghram is the meaning.The anvaya or relations between tu or atha and other words are of four typesdepending on the category of the other word 1) noun 2) gender 3) pronoun4) indeclinable.Noun word (1.1.45 a.k.)The word tu which is connects with the word amarvat here and not withthe word nagar. (2.8.73 a.k.)Here the word atha connects with the word tvaritam and not with the wordghra.Gender word (1.3.12 a.k.)Here tvanta pusi (pusi the word that ends with tu) connects withantardhi. (1.4.26 a.k.)Here athdi trisu (the word trisu which starts with atha) connects withdravye. Pusi and trisu these two gender informations will go to the wordsantardhi and dravye.Pronoun word (1.1.44 a.k.)Here tasya the pronoun will connect with the word priy.42Indeclinable word Here v the indeclinable will connect with the gender informating wordpusi.The atha indicates atho which is the synonym of atha also. E.g. :- (1.10.19 a.k.)4.2 LigdisagrahavargaThe last chapter of Amarakoa named as ligdisagrahavarga. Amarasimhaagain explains the rules regarding gender in detail. These rules maybe divided into eight parts. Those are strligaprakaraam (rulesfor feminine gendered words), pulligaprakaraam (rules for masculinegendered words), napusakaligaprakaraam (rules for neuter genderedwords), punapusakaligaprakaraam (rules for masculine-neuter genderedwords), strpulligaprakaraam (rules for feminine-masculine genderedwords), strnapusakaligaprakaraam (rules for feminine-neuter genderedwords), triligaprakaraam (rules for feminine-masculine-neuter genderedwords), and paravalligaprakaraam (rules for the gender of thecompound-ending-words). Some suxes indicate only a particular gender,so he lists those suxes as well. Sometimes he lists the words themselvesparticularly when these words can not be captured under any generalities.Sometimes he declares that synonyms of the words are also in the samegender only. E.g. :- (3.5.3 a.k.)The words vidyut, ni, vall, v, dik, bhmi, nad, hr and their synonymsare in feminine gender. adantaikrthadvigu are feminine E.g. pacaml,packsar etc. But not in the case of ptra, yuga etc. E.g. pacaptram,cauryugam, tribhuvanam, etc.So we can cross check the gender information of the words which was listedby Amarasimha , using ligdisagrahavarga.43Commentaries of Amarakoa are very much useful to decide the perfectgender when there is a confusion. Vykhysudh or Rmram were givenrst preference in case of conict. The Malayalam Commentaries basedon Sudha named as pramevar and trive were also useful. The otherCommentaries like ksarvasvam, Amarakoodghanam and Mahevarakalso referred to when in doubt. After doing the gender marking according tothe rules given by Amarasimha, we did a cross check with Devadatta TivarisDevakoa artht Amarakoa and Colebrookes commentary on Amarakoa .4.3 Comparison with Monier WilliamsDictionaryWe did a comparison of the gender of words in Amarakoa with the gender ofthose words as mentioned in Monier Williams Sanskrit-English Dictionary.The comparison was done automatically by through some simple "perl"scripts.4.3.1 Monier Williams Sanskrit - English DictionaryIt is a Sanskrit English Dictionary developed by Sir Monier MonierWilliams (MW), rst published by Oxford University Press in 1899. He wasthe second occupant of the Boden Chair of Sanskrit at Oxford University.It is an autentative modern Sanskrit dictionary. It's another advantage is,it is available in electronic form.We did a comparison between the gender information given by Amarasimhawith MW's gender information. The purpose of this exercise was to see howmuch deviation is there as regards the gender, over a period of centuries.4.3.2 The Result of The ComparisonTotal tokens in Amarakoa are 11,580 and among these 5886 words had thesame gender in MW. The other words, where there was a deviation fall intothree categories - partial match, extra gender or total mismatch.44Partial MatchThese are the cases where each Amarakoa as well as MW report morethan one genders for a word, but the genders in both the lexicon do notmatch totally. They match only partially. E.g. :- the Prtipadika "garta"has masculine and feminine gender in Amarakoa and masculine and neutergender in MW. So here masculine is a common gender and in Amarakoafeminine is an extra gender, not found in MW and in MW neuter gender isan extra gender not found in Amarakoa. Figure 4.1 shows the Venn diagramof the partial match of the gender in Amarakoa and MW.Figure 4.1: Partial match of Amarakoa and MWOnly three such words were found. For all these words Amarasimhaclearly mentions the gender and hence there is even no doubt about theinterpretation of their gender.1. The prtipadika "kroa" is in feminine and neuter in Amarakoaaccording to the verse - (2.6.77 a.k.)But it is in masculine and neuter gender in MW3. In Amarakoa theword n is used to denote masculine gender. Here Amarasimha negates3MW. p.32345the masculine gender by na n. Negation indicates the other twogenders, so the word is in feminine and neuter genders.2. The prtipadika "viva" is in feminine and neuter genders in Amarakoaaccording to the verse - (2.9.38 a.k.)But in MW it is in masculine and feminine gender4.3. The prtipadika "garta" is in masculine and feminine genders accordingto the Sudhvykhy of Amarakoa. Sudhkra quotes Rabhasa andHaima, the Sanskrit lexicographers5. But in MW it is marked asmasculine and neuter genders6.Extra Gender in Amarakoa or in MWIn some cases either Amarakoa or MW has an extra gender over and abovewhat the other lexicon includes. Like these more than 5000 words are therein Amarakoa. Figure 4.2 and gure 4.3 represents Venn diagrams of theinclusion.This clearly indicates the diachronic change in the value of the gender.Mismatches of Word Genders in Amarakoa and in MWMismatches of the genders between Amarakoa and MW are moreimportant than inclusion or intersection of genders. In the Sanskrit word'sgender informations, there can be variations according to lexiconsor lignusanastram or modern type dictionaries. Lexicons orlignusanastram are representing the gender which was used inancient times. There can be changes in gender according to their dierenttype usages in modern times, because usages will change according to time.So it is very much important to watch how it is diered.In some cases, the verses of Amarakoa itself provides gender informationusing some abbreviations, but in these types of case also MW is not matching4MW. p.9925Vykhysudh p.87.6MW. p.34946Figure 4.2: Amarakoa gender included in to MWFigure 4.3: MW gender included in to Amarakoawith this some times. So according to Amarasimha those are correct. Thesetypes of cases are ve in total Amarakoa . List is given in Appendix-B.E.g.:- The prtipadika "srasana" is in feminine gender according to the verse-47 (2.6.109 a.k.)Here Amarasimha indicating it as neuter using the word klbe. But in MWit is marked as masculine gender7.Lignusanastra's are also helping to decide gender when it has confusinggender. Sudh itself describes about this stras when it is needed. So fromthis also we can decide gender.E.g.:-The prtipadika "chadis" is in feminine gender according to theLignusanastra 8and Amarakoa Commentaries are alsomarking it as feminine only. But in MW it's gender is recorded as neuter9.Some times the gender of the word can be decided from verses through it'sword endings. But, some of these type of cases are also not matching withMW. These types of cases from Amarakoa are given in Appendix-B.E.g.:- ajagava is the prtipadika in the meaning of the 'bow of god iva'.Amarakoa marks it as neuter gender through the verse - (1.1.35 a.k.)But in MW it is marked as masculine gender10.For some cases we can decide the word gender from the compound ending.Because there is a rule in Pini's vykaraa that 11

It tells that the compounds in dvandva and in tatpurusa will take the genderof the last word. So we can decide the last word's gender in the dvandvaand tatpurusa compound. Some of these type of cases are also not matchingwith MW. These types of cases are ve in Amarakoa .1. The prtipadika "dhaivata" is masculine gender marked as Amarakoa . (1.1.7 a.k.)7M.W p.12098 - 219MW. p.40410MW. p.911 2/4/2648The dvandva compound ending is in masculine gender here, andthe ending word is dhaivata. So we can decide that the prtipadika"dhaivata" is in masculine gender. But in this case MW marked it asneuter gender12.2. The prtipadika "tejana" is in masculine gender in Amarakoa accordingto the verse - (2.4.161 a.k.)Here this is a dvandva compound ending in masculine gender,and the ending word is tejana also. According to Pini the genderof tejana will be masculine. But in MW it is marked as neuter gender13.3. The prtipadika "sayuga" is in masculine gender in Amarakoaaccording to the verse - (2.8.105 a.k.)Here this is a dvandva compound ending in masculine gender, andthe ending word is sayuga also. According to Pini the gender ofsayuga will be masculine. But in MW it is marked as neuter gender14.4. The prtipadika "rauhisa" is in neuter gender in Amarakoa accordingto the verse - (2.4.166 a.k.)Here this is a dvandva compound ending in neuter gender, and theending word is rauhisa also. According to Pini the gender of rauhisawill be neuter. But in MW it is marked as masculine gender15.12MW. p.52013MW. p.45414MW. p.111215MW. p.891495. The prtipadika "bhsta" is in neuter gender in Amarakoa accordingto the verse - (2.4.167 a.k.)Here this is a dvandva compound ending in neuter gender, andthe ending word is bhsta also. According to Pini the gender ofbhsta will be neuter. But in MW it is marked as masculine gender16.In some cases it is very dicult to decide the gender of compound words inAmarakoa. E.g. :- (1.1.6 a.k.)Deciding the gender of the word tridiva from this compound is very muchdicult. The word is in the middle of the coordinative compound. In thesetypes of cases we are following the information from the commentaries. Forthe word tridiva, commentaries are marking it as in masculine gender. ButMW is marking as neuter and feminine gender17. These types of cases arelisted in Appendix-B.4.4 Polysemy Distribution in AmarakoaAmarakoa has 4,053 synsets. Some of the words fall under more thanone synsets, and thus are ambiguous. Most of these polysemous wordsbelong to the nnrthavarga of the third ka which lists the polysemouswords alphabetically according to their endings like knta, khnta, gntaetc. nnrthavarga has 814 polysemous words. In Amarakoa, 21% (2545)words have more than one meaning. The polysemy distribution in theAmarakoa is summarized in Table 4.1. There is only one word hari inAmarakoa which has as many as 14 senses, the word antar belongs to 13synsets, and the word go has 12 synsets. We note that almost 65% words(7459 words) belong to a single synset and thus are not ambiguous.16MW. p.76117MW. p.45850No. of meanings No. of words Words14 1 hari13 1 antar12 1 go10 2 kri, ka9 2 rasa, vsa8 8 dhtu, dharma, vasu, arisa...7 96 185 494 1363 3302 10151 7459Table 4.1: Polysemy Distribution51Chapter 5Knowledge Structure inAmarakoaThe Indian tradition of transmitting knowledge orally is on the verge ofvanishing. As the oral transmission demands, Indian traditional educationalculture was organised to be formal and intensive as opposed to themodern culture which is more informal and extensive (Wood, 1985). Intraditional circumstances, a child would receive his education largely byoral transmission, mainly through rote-learning. The method employedwas through recitation and remembering. A child is taught the alphabet(varaml), he would memorise a few verses, subhsitas, and then startreciting a dictionary of synonymous words -- the Amarakoa -- till it ismemorised. It typically would take anywhere between 6 months to a yearto memorise a list of approximately 10,000 Sanskrit words arranged as a listof synonyms. The close inspection of the structure of the Amarakoa givesmuch more insight into the way the words are organised. When a studentmemorises it, though in the beginning it appears as a linear list of words, ashe starts understanding the meaning of the words, reads the commentarieson this text and starts using these words, the linear structure unfolds into aknowledge web with various links.The Amarakoa printed in the form of a book just shows the linear order, andthe index at the end of the book point to various words for easy references.But there is much more to it than just a linear order. The knowledge astudent acquires through various commentaries and also its practical usein his own eld of expertise -- be it yurveda, vykaraa or shitya, is52in the form of various links. With the modern education culture that isdominated by the use of computers as a tool, which relies more on thesecondary memories such as books, computers, and the World Wide Web,than the human memory, it is necessary to make the implicit knowledgein Amarakoa explicit. The computers have an advantage over the printedbooks. Computers can represent multi-dimensional objects, and thus onecan navigate through the whole structure and at the same time with thepowerful search facilities can search complex queries. Here in this chapterwe present a computational model that, can 'visualise' various kinds of linksin Amarakoa, and provide a database model to store these links in order tofacilitate automatic extraction of these links as an answer to a search query.5.1 AmarakoaAs explained in the rst chapter Amarakoa consists of 1068 lokas. Theseare divided into three kas and each ka is divided in to vargas. Thereare 25 vargas in total. Amarakoa contains 11,580 content words (tokens).Some of the tokens are repeated either within a ka or across the kasleading to only 9,031 types.Amarakoa as is well known consists of paryyapada's (synonymouswords). A set of synonymous words is termed as a synset. Each synonymmay span over one or more verses. The following verse, e.g., provides asynonym for the word jambuka. (2.5.5 a.k.)5.2 Organisation of synsets within a vargaExcept the polysemous words (nnrthavarga), all other synsets in a vargashow some semantic relation to the varga it belongs to and sometimes even tothe preceding or following synsets. These semantic relations indicate variouskinds of relations. They may be classied as hierarchical or associative.The hypernym indicating a more general term or the hyponym showinga more specic term are the examples of hierarchical relation. Similarlythe holonym-meronym relation marking the whole-part relation is also a53hierarchical relation. In addition various other relations are indicated bythe adjacency of the synsets. These may be termed as associative relations,which indicate some kind of association of one synset with the other. Thisassociation may be the association among human beings, or the associationof certain objects with certain other objects. We illustrate below some suchrelations with examples.5.2.1 Example 1: VisuhThe verses from 1.1.18 to 1.1.29 describe various synsets representing visu,and objects related to/associated with visu. The relations, as is evidentfrom the following description, are kinship relations such as father, brother,son, grandson, wife, and also associated objects such as conch, discus, sword,vehicle, etc. (See Figure 5.1).visu (1.1.18 - 1.1.22)1ksa's father (1.1.22)ksa's elder brother (1.1.23 - 1.1.24)kmadeva (1.1.25 - 1.1.26)oral arrows of kmadeva (1.1.26)physical arrows of kmadeva (1.1.26)son of kmadeva - aniruddha (1.1.27)wife of visu - laksm (1.1.27)Special devices/equipments of visu (1.1.28)(conch, discus, sword, jewel, bow, horse, mark,etc.)ksa's charioteer, minister (1.1.28)ksa's younger brother (1.1.28)visu's vehicle - garua (1.1.29)5.2.2 Example 2: SamayaThe verses from 1.4.1 to 1.4.9 deal with words related to time, units ofmeasurement, special names of special days, etc.1The English translations of the subheadings, which are given here and in the followingexamples, describing the lokas are taken from Colebrooke's commentary on Amarakoa(Colebrooke, 1808).54Figure 5.1: Relations of visuTime (1.4.1)Lunar day (1.4.1)First lunar day (1.4.1){Day (1.4.2)Morning (1.4.2 - 1.4.3)Twilight (1.4.3)Evening (1.4.3)First four hours of a day (1.4.3)Second four hours of a day (1.4.3)Third four hours of a day (1.4.3)Period of the day (1.4.3)Night (1.4.3 - 1.4.4)A dark night (1.4.5)A moonlit night (1.4.5)A night and two days (1.4.5)First part of night (1.4.6)Midnight (1.4.6)Sequence of nights (1.4.6)Space of three hours (1.4.6) }Last day of the half month (1.4.7)Precise moment of the full or the new moon (1.4.7)Full moon day (1.4.7)Full moon whole day(1.4.8)Full Moon with a little gibbous on part of a day (1.4.8)55No moon day (1.4.8)waning crescent (1.4.9)No moon whole day (1.4.9)In this example we also see violation of nesting. In between the synsetsrelated to lunar day and last day of the month, the synsets related to day(which refers to the apparent solar motion) are intervened.5.2.3 Example 3: KsatriyaHere is a group of verses from 2.8.1 to 2.8.10 belonging to the Ksatriyavarga.The words here refer to the king, military, sministers, various category ofpeople engaged in the services of kings, etc.Man of the military tribe (2.8.1)King (2.8.1)An emperor (2.8.2)Universal monarch (2.8.2)King over a country (2.8.2)Paramount sovereign (2.8.3)Multitude of kings (2.8.3)Multitude of military tribe (2.8.4)Minister (2.8.4)Deputy minister (2.8.4)Priest (2.8.5)Judge (2.8.5)King's companions (2.8.5)Body guards of a king (2.8.6)Warder (2.8.6)Superintendent (2.8.6)Village Superintendent (2.8.7)Superintendent of many villages (2.8.7)Superintendent of Gold (2.8.7)Superintendent of Silver (2.8.7)Superintendent of the womens' apartments (2.8.8)Outside guard of the womens' apartment (2.8.8)attendant of a king (2.8.9)eunuch (2.8.9)56Prince whose territories lie on the frontiers of those of the enemy(2.8.9)Neighboring prince (2.8.9)Prince whose territories lie beyond those of the friend (2.8.10)Enemy in the rear (2.8.10)5.3 Implicit relationsThese were three samples from three distinct topics involving totally dierentkind of relations. All these relations are semantic in nature. A more detailedstudy of such examples showed that following relations occur more frequently. avayavvayav (part-whole relation) parparjti (is a kind of relation) janyajanaka (child-parent relation) patipatn (husband-wife relation) svasvmi (master-possession relation) jvik (livelihood)5.3.1 Is a part of (avayavvayav)Syn()2= , , , , , , , , ,, , , , , , .andSyn() = , . , are part of , , , etc.. Hence is marked tobe is_a_part_of ( of) Similarly , ( Syn ()) are also part of , , ,etc.. Hence , where Syn ()= , 2synset of 57also bears a part_of relation with . See the Figure 5.2.Figure 5.2: Part-Whole Relation5.3.2 Is a kind of (parparjti)Syn() = , , , , , , , ,, , , , , , , , , ,.andSyn() = , , , , , , , ., , ,....... are a kind of , , , ........., etc..Hence is marked to be is_a_kind_of . See the Figure 5.3.Figure 5.3: Is-a-kind-of Relation5.3.3 Child-parent relation (janyajanakabhva)Syn () = , .andSyn () = , , , , , , , ,, , , , , , , , , , ,58, , , , , , , , , , , , , , , , , , , .The head-words and shares the janyajanakabhva (parent-childrelation). See the Figure 5.4.Figure 5.4: Parent-child Relation5.3.4 Husband-wife relation (patipatnbhva)Syn() = , , , , , , ,, , , , , , , .andSyn( ) = , , , , , , , , ,, , , , , , , , , , , , , , , , , ,, , , , , , , ,, , , , , , , ,, , , , , , , , , , , .Here the head-word laksm has the husband relation with the head-wordvisuh and visuh has the wife relation with laksm. See the Figure 5.5.5.3.5 Master-possession relation (svasvmibhva)Syn( ) = ,59Figure 5.5: Husband-Wife RelationSyn( ) = ,Syn() = , , , , , , , , , andSyn( ) = , , , , , , , , ,, , , , , , , , , , , , , , , , , ,, , , , , , , ,, , , , , , , ,, , , , , , , , , , , .Here the head-word visuh has the master relation with the head-words , and , and vise versa these three have thepossession relation with the head-word visuh. See the Figure 5.6Figure 5.6: Master-possession Relation605.3.6 Livelihood (jvik)The synset with head-word matsya is (, , , , , , , , ) which denotes objects which act as a livelihoodfor the objects expressed through the concept of dhvara, and hence thelivelihood for the objects belonging to the synset dhvara is marked as amatsya. (see Figure 5.7).Figure 5.7: Livelihood RelationFor more examples, see the Appendix B.There are a few other relations such as kinship relations, dhrdheya,vaavaya etc. But the instances of such relations were found to be rare.61Chapter 6Ontological Representationscheme for Amarakoa'Ontology' the word with Greek origin means "study of being or existenceor reality". The study of existence of entities, their classication based onsimilarities or dierences, the hierarchy involved in it come more close to thephilosophical studies. Ontology also plays an important role in KnowledgeRepresentation.Knowledge Representation is a way of specication or representation of aconcept. According to J. F. Sowa, it is a multidisciplinary subject thatapplies theories and techniques from three other elds: Logic, Computation,and Ontology.Logic provides the formal structure and rules of inference.Computation supports the applications that distinguish knowledgerepresentation from pure philosophy.Ontology denes the kind of things that exist in the application domain1.We can illustrate knowledge representation like this :- if we considerknowledge representation as a house construction, then the sketch ofthe house is provided by logic, construction works are undertaken by1J.F Sowa,2000 Preface,XI,XII.62computation, and the materials for the house are supplied by ontology.Without the materials, using only sketch and construction works, we cannotcreate a house. Like that without the knowledge of existence, the logic andthe computation cannot fulll the knowledge representation of a concept.6.1 OntologyOntology constitutes the major branch of philosophy viz. Metaphysics2.J.F Sowa explains metaphysics like this:- "The two sources of ontologicalcategories are observation and reasoning. Observation provides knowledge ofthe physical world, and reasoning makes sense of observation by generatinga framework of abstractions called metaphysics3." Later the term 'ontology'has been adopted by the Articial Intelligence research into knowledgerepresentation. The AI school deals with the organisation of knowledgeabout the universe. In order to provide a suitable structure to the knowledge,AI researchers need an ontology for the domain taken for application."The subject of ontology is the study of the categories of things that existor may exist in some domain"4."A choice of ontological categories is the rst step in designing a database,a knowledge base, or an object-oriented system. In database theory thecategories are usually called domains, in AI".6.1.1 Denition of OntologyWordNet lists two senses for the term "ontology". The rst is given accordingto the computer science point of view and the second is according to thephilosophy point of view.1. "a rigorous and exhaustive organization of some knowledge domainthat is usually hierarchical and contains all the relevant entities and2Metaphysics is a branch of philosophy concerned with explaining the fundamentalnature of being and the world.3J.F Sowa,2000 Pg.No.514J.F.Sowa,2000 Pg.No.49263their relations", and the hypernymy relation shows like this5:-=> arrangement, organization, organisation, system=> structure=> cognition, knowledge, noesis=> psychological feature=> abstraction, abstract entity=> entity2. "the metaphysical study of the nature of being and existence", and thehypernymy relation shows like this :-=> metaphysics=> philosophy=> humanistic discipline, humanities, liberal arts, arts=> discipline, subject, subject area, etc.=> know