Extracting Context

download Extracting Context

If you can't read please download the document

  • date post

  • Category


  • view

  • download


Embed Size (px)

Transcript of Extracting Context

Slide 1

01/10/2010Extracting meaning from Social Media Monitoring.OR


Is they often present you with this...01/10/20103

Understanding the meaning of results and how to use monitoring tools so they work for you Filtering the informationUsing a top down approach to filtering often wastes time, money and effortTry a bottom up approach to define filters - its quicker and it will help understand the nature of the social media and how your brand interacts with it better - particularly useful when you need to expand the searchUse as many specifics as you can ( that may be none, but usually isnt)Be prepared to spend some time planning how you will use filters to categorize and quantify data (every hour you spend at the planning stage can save you days in execution - and can be especially valuable when you are briefing vendors)Make sure the filters make sense to your business and monitoring objectives 01/10/20104


The perfect tool?

How much do you already know about your audience?Take the time to establish which social media channels it makes sense for your audience to use

Better still, choose a tool which allows you to check your assumptions

Experiment and test your assumptions before you go setting up a whole lot of details in specifications and filteringUse exceptions to exclude unwanted audiencesMake sure you understand the vocabulary your audiences use 01/10/20106

01/10/20107Data CleanupThe problem with social media is that finding 20,000 hits is a lot easier than finding the 50 that matterFilters and exclusions can screen out obvious mismatches but can also cause you to miss important itemsA successful strategy for cleaning data without losing integrity should include:Contextual matches rather than keywordsNon-linguistic cues and vocabularyLinks and associations (especially where you can set limits/minimum values)Use of active terms Use of specialized tags and terms to allow easier identification of who is responding to your brands contentIncorporation of SEO terms used in materials designed for audiencesLinks outside social media groups for source and destination tagging

Customizing the sampleThe first rule of data sampling should be: design backwards from the intended result/audience for your resultsEstablish what you are going to do with results Who are the ultimate recipientsDo results have be consistent with other business functions resultsDont generate information no one can use ( i.e. if you are monitoring in real time - make sure there are real time processes which uses these resultsQuestion outside your own immediate needs are you monitoring for research, evaluation, discovery, actionable insight, immediate response?Make sure your cleaned data can be transformed into all the outlets and reports you require


A matter of PerspectivePerspectives are a quick and easy way of overcoming one of the most persistent problems with analyzers - making the answer relevant to you.In data cleanup they can be used to eliminate a lot of irrelevant contentIn research they can be used to shift the view between , say, competing brands or topicsMost analysers use people to add the perspective.......but there is an alternative

Automated perspectives can be specified as part of the filtering and cleanup processstored perspectives allow a rapid comparison of data from different viewpoints which multiplies the business/media intelligence in the sample Anything automated can be switched on or off and can be modifiedConsistency in results will go up01/10/20109

ContextMost sentiment analysers do not allow or imply context in their resultswhich is why the results often dont look right Context uses perspectives and filters to provide a means of matching the results to the brand or person relating to the monitoring objectivePut simply - the results should make sense Context also allows the focus of monitoring to reflect actual process in a business e.g. A buy decision or an expression of discontent about a serviceThe role of context is to make You the focus of the answer01/10/201010

Framing QuestionsThe most important rule is to make sure the answers are answering YOUR questions Either make sure you test logical questions ahead of time


An end to Keywords?Look at the new generation of analyzers which combine natural language and context to generate results which mean what you said in your questionThe benefit is that missed hits are minimised simply because your vocabulary missed a word or variation, or the time tense is different, or one or more of the key elements isnt language (such as an emoticon, or slang expressionsThis approach allows you to work in a more natural way in both framing questions and evaluating outcomes01/10/201012

The trouble with automated sentiment analysisIs often what it is expected to accomplishIf you have designed in the filtering, contextual mapping and the end user of the information it is perfectly practical to expect a result in the 90-95% accuracy rangeBeware of claims over 95% (in a live language, subject to fashion and the re-use of terms in new ways there are very real and tangible barriersBut dont make the mistake of assuming a human reader can do any better - a 2009 study of 2,000 people saw them score an average of 84% - and 88% was the top markThere is a semantic trap in most analysers in that they use language as it is defined - not as it is actually used - and most rely on words (social media posters sometimes dont) - a lot of language isnt traditional linguistics!Black box solutions, however good they appear, are hiding things from you - insist on transparency 01/10/201013

Death to all humans?01/10/201014

Why automated analysis is NOT about replacing peopleThe role of automation is not about replacing people from the process rather is should be about allowing you more time to think about what is important rather than speed readingTake the Apple iPhone 4 - social media DISCUSSION THREADS not posts were ruining at 30/second during the launch day - leaving anyone trying to read and make sense of the sentiment in a passive, or reactive state.This is an extreme, but most social media sites can easily overwhelm a human-based system allowing people to do little more than skim contentContext- based analysis can reduce the results to actions and insights which are both manageable and insightful - without increasing the errors from volume - without getting tired and it will read every blog or post all the way through.


Making results countAVE? ROI? If you are going to be subject to externally generated measures (often for historical or consistency reasons) make sure you can measure something meaningful for the purposeBetter still develop a metric which actually makes sense for what you are doing and communicate its benefitsMake sure your results are not siloed to just social media - there are often links to other media forms (including your own materials) make sure your solution integrates them all and can show you these relationshipsDashboards:Employ ways of making your results and their context stand out dashboards are both an excellent way of compressing complex information and adding impact to results (but too often are generalized templates presented by vendors make sure you get what you want how you want it)Can be used to extend your information into other parts of the businessCan create synergy with other business processes - make sure your social media dashboard can take on information from other functionsDont misuse dashboards to create meaningless measurements or introduce concepts which make no sense to anyone other than the graphics/metrics person who designed it - stick to what you know and can understand01/10/201016

And finally....

Feedback loops both in the human and automated sense are a valuable mechanism to shrink wrap results ever closer to you brand objectives - use them!01/10/201017