Predictive Text Analytics
Download as PPT, PDF6 likes3,756 views
The document discusses predictive text analytics, including predicting text completions, disambiguating text, and correcting errors. It also discusses extracting entities, concepts, facts, and sentiments from unstructured text sources for applications like search, knowledge discovery, and predictive analytics. Key challenges include the complexity of human language with features like ambiguity and context.
Ad
More Related Content
What's hot (20)
Viewers also liked (8)
Ad
Similar to Predictive Text Analytics (20)
Ad
More from Seth Grimes (20)
Recently uploaded (20)
Predictive Text Analytics
- 1. Predictive Text Analytics Seth Grimes Alta Plana Corporation 301-270-0795 -- http://altaplana.com -- @SethGrimes Predictive Analytics World October 20, 2009
- 3. What is Analytics? http://www.tropicalisland.de/NYC_New_York_Brooklyn_Bridge_from_World_Trade_Center_b.jpg x(t) = t y(t) = ½ a (e t/a + e -t/a ) = acosh(t/a) http://en.wikipedia.org/wiki/Seven_Bridges_of_K%C3%B6nigsberg c:\temp\Marcus.pdf c:\temp\Marcus.pdf
- 4. What is Analytics? "SUMLEV","STATE","COUNTY","STNAME","CTYNAME","YEAR","POPESTIMATE", 50,19,1,"Iowa","Adair County",1,8243,4036,4207,446,225,221,994,509 50,19,1,"Iowa","Adair County",2,8243,4036,4207,446,225,221,994,509 50,19,1,"Iowa","Adair County",3,8212,4020,4192,442,222,220,987,505 50,19,1,"Iowa","Adair County",4,8095,3967,4128,432,208,224,935,488 50,19,1,"Iowa","Adair County",5,8003,3924,4079,405,186,219,928,495 50,19,1,"Iowa","Adair County",6,7961,3892,4069,384,183,201,907,472 50,19,1,"Iowa","Adair County",7,7875,3855,4020,366,179,187,871,454 50,19,1,"Iowa","Adair County",8,7795,3817,3978,343,162,181,841,439 50,19,1,"Iowa","Adair County",9,7714,3777,3937,338,159,179,805,417
- 5. www.stanford.edu/%7ernusse/wntwindow.html Axin and Frat1 interact with dvl and GSK, bridging Dvl to GSK in Wnt-mediated regulation of LEF-1. Wnt proteins transduce their signals through dishevelled (Dvl) proteins to inhibit glycogen synthase kinase 3beta (GSK), leading to the accumulation of cytosolic beta-catenin and activation of TCF/LEF-1 transcription factors. To understand the mechanism by which Dvl acts through GSK to regulate LEF-1, we investigated the roles of Axin and Frat1 in Wnt-mediated activation of LEF-1 in mammalian cells. We found that Dvl interacts with Axin and with Frat1, both of which interact with GSK. Similarly, the Frat1 homolog GBP binds Xenopus Dishevelled in an interaction that requires GSK. We also found that Dvl, Axin and GSK can form a ternary complex bridged by Axin, and that Frat1 can be recruited into this complex probably by Dvl. The observation that the Dvl-binding domain of either Frat1 or Axin was able to inhibit Wnt-1-induced LEF-1 activation suggests that the interactions between Dvl and Axin and between Dvl and Frat may be important for this signaling pathway. Furthermore, Wnt-1 appeared to promote the disintegration of the Frat1-Dvl-GSK-Axin complex, resulting in the dissociation of GSK from Axin. Thus, formation of the quaternary complex may be an important step in Wnt signaling, by which Dvl recruits Frat1, leading to Frat1-mediated dissociation of GSK from Axin. www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Retrieve&list_uids=10428961&dopt=Abstract The “Unstructured Data” Challenge
- 6. Modelling Text Metadata E.g., title, author, date Statistics Typically via vector space methods E.g., term frequency, cooccurrence, proximity Linguistics Lexicons, gazetteers, phrase books Word morphology, parts of speech, syntactic rules Larger-scale structure including discourse Machine learning
- 7. “ Statistical information derived from word frequency and distribution is used by the machine to compute a relative measure of significance, first for individual words and then for sentences. Sentences scoring highest in significance are extracted and printed out to become the auto-abstract.” H.P. Luhn, The Automatic Creation of Literature Abstracts , IBM Journal , 1958.
- 8. Text Modelling Document content can be considered an unordered “bag of words.” Particular documents are points in a high-dimensional vector space. Salton, Wong & Yang, “A Vector Space Model for Automatic Indexing,” November 1975.
- 9. Text Modelling We might construct a term- document matrix ... D1 = "I like databases" D2 = "I hate hate databases" and use a weighting such as TF-IDF (term frequency–inverse document frequency)… in computing the cosine of the angle between weighted doc-vectors to determine similarity. http://en.wikipedia.org/wiki/Term-document_matrix I like hate databases D1 1 1 0 1 D2 1 0 2 1
- 10. Text Modelling Analytical methods make text tractable. Latent semantic indexing utilizing singular value decomposition for term reduction / feature selection. Creates a new, reduced concept space. Takes care of synonymy, polysemy, stemming, etc. Classification technologies / methods: Naive Bayes. Support Vector Machine. K-nearest neighbor.
- 11. Text Modelling In the form of query-document similarity , this is Information Retrieval 101. See, for instance, Salton & Buckley, “Term-Weighting Approaches in Automatic Text Retrieval,” 1988. A useful basic tech paper: Russ Albright, SAS, “Taming Text with the SVD,” 2004. Given the complexity of human language, statistical models may fall short.
- 12. Semantics “ This rather unsophisticated argument on ‘significance’ avoids such linguistic implications as grammar and syntax... No attention is paid to the logical and semantic relationships the author has established.” -- Hans Peter Luhn, 1958
- 13. New York Times , September 8, 1957 Anaphora / coreference: “They”
- 14. Semantics Why do we need linguistics? The Dow fell 46.58, or 0.42 percent, to 11,002.14. The Standard & Poor's 500 index fell 1.44, or 0.11 percent, to 1,263.85, and the Nasdaq composite gained 6.84, or 0.32 percent, to 2,162.78. The Dow gained 46.58, or 0.42 percent, to 11,002.14. The Standard & Poor's 500 index fell 1.44, or 0.11 percent, to 1,263.85, and the Nasdaq composite fell 6.84, or 0.32 percent, to 2,162.78. (Example: Luca Scagliarini, Expert System.) John pushed Max . He fell. John pushed Max. He laughed. (Example: Laure Vieu and Patrick Saint-Dizier .)
- 21. Text Analytics Text analytics automates what researchers, writers, scholars, and all the rest of us have been doing for years. Text analytics -- Applies linguistic and/or statistical techniques to extract concepts and patterns that can be applied to categorize and classify documents, audio, video, images. Transforms “unstructured” information into data for application of traditional analysis techniques. Unlocks meaning and relationships in large volumes of information that were previously unprocessable by computer. ... models text .
- 22. Predictive Text Analytics? Let’s consider three interpretations: Predictive text analytics Prediction applied to text. Predictive analytics from text sources Analysis of information extracted from text. Predictive text analytics Clustering and classification of the text at document & feature levels.
- 23. Predictive Text Analytics Predictive text analytics Prediction applied to text. Predictive analytics from text sources Analysis of information extracted from text. Predictive text analytics Clustering and classification of the text at document & feature levels.
- 24. Predictive Text Basic modelling to facilitate functions such as: Completion Disambiguation : use dictionaries, context Error correction http://en.wikipedia.org/wiki/File:ITap_on_Motorola_C350.jpg
- 25. Predictive Text Marti Hearst in Search User Interfaces : “ Search logs suggest that from 10-15% of queries contain spelling or typographical errors. Fittingly, one important query reformulation tool is spelling suggestions or corrections.”
- 26. Predictive Text Analytics Predictive text analytics Prediction applied to text. Predictive analytics from text sources Analysis of information extracted from text. Predictive text analytics Clustering and classification of the text at document & feature levels.
- 27. “ The bulk of information value is perceived as coming from data in relational tables. The reason is that data that is structured is easy to mine and analyze.” – Prabhakar Raghavan, Yahoo Research, former CTO of enterprise-search vendor Verity (now part of Autonomy) , 2004 Yet it’s a truism that 80% of enterprise information is in “unstructured” form. Predictive Analytics, Text Sources
- 28. Consider: E-mail, news & blog articles, microblogging, forum postings, and other social media. Contact-center notes and transcripts; recorded conversations. Surveys, feedback forms, warranty claims, case reports. And every other sort of document imaginable. These sources may contain “traditional” data. The Dow fell 46.58, or 0.42 percent, to 11,002.14. The Standard & Poor's 500 index fell 1.44, or 0.11 percent, to 1,263.85, and the Nasdaq composite gained 6.84, or 0.32 percent, to 2,162.78. Sources
- 29. Search++ Search is typically answer #1.
- 30. Beyond Search Text analytics extracts and classifies by – Entities: names, e-mail addresses, phone numbers Concepts: abstractions of entities. Facts and relationships. Events. Abstract attributes, e.g., “expensive,” “comfortable” Opinions, sentiments: attitudinal data. – for search indexes, knowledge bases, and databases.
- 33. Applications Text analytics has applications in – Intelligence & law enforcement. Life sciences. Media & publishing including social-media analysis. Competitive intelligence. Voice of the Customer: CRM, product management & marketing. Legal, tax & regulatory, compliance. HR & recruiting. Great lift potential when coupled with transactional & operational data.
- 34. Predictive Text Analytics Predictive text analytics Prediction applied to text. Predictive analytics from text sources Analysis of information extracted from text. Predictive text analytics Clustering and classification of the text at document & feature levels.
- 35. Document processing -- This slide and the next show dynamic, clustered search results from Grokker (now gone)…
- 36. … with a zoomable display. Clustering here identifies cohesive groupings of retrieved documents.
- 38. Sentiment Analysis “ Sentiment analysis is the task of identifying positive and negative opinions, emotions, and evaluations.” -- Wilson, Wiebe & Hoffman, 2005, “ Recognizing Contextual Polarity in Phrase-Level Sentiment Analysis” Steps include: 1) detection, 2) classification, 3) measurement: WW&H (for example) used over 8,000 subjectivity indicators. Polarity: positive, negative, (both,) or neutral. Intensity.
- 39. Complications. Levels: Corpus / data space, i.e., across multiple sources. Document. Statement / sentence. Entity / topic / concept. Language characteristics: Jargon, slang, irony, ambiguity, anaphora, polysemy, synonymy, etc. Context is key. Discourse analysis comes into play. Sentiment holder ≠ object: Geithner said unemployment will worsen… Sentiment Analysis
- 40. Steps in the Right Direction
- 42. Unfiltered duplicates External reference “ Kind” = type, variety, not sentiment. Naïve misclassification ... and Missteps
- 43. “ We present a system that adds an emotional dimension to an activity that Internet users engage in frequently, search..” -- Sood & Vasserman & Hoffman, 2009, “ ESSE : Exploring Mood on the Web” Beyond Polarity
- 44. Happy Sad Angry Energetic Confused Aggravated Bouncy Crappy Angry Happy Crushed Bitchy Hyper Depressed Enraged Cheerful Distressed Infuriated Ecstatic Envious Irate Excited Gloomy Pissed off Jubilant Guilty Giddy Intimidated Giggly Jealous Lonely Rejected Sad Scared ----------------------- The three prominent mood groups that emerged from K-Means Clustering on the set of LiveJournal mood labels.
- 45. Predictive Text Analytics Predictive text analytics Prediction applied to text. Predictive analytics from text sources Analysis of information extracted from text. Predictive text analytics Clustering and classification of the text at document & feature levels.
- 46. Seth Grimes Alta Plana Corporation 301-270-0795 – http://altaplana.com @SethGrimes
Editor's Notes
- #33: This course is, in essence, about the information enterprises have and how they use it and how they could better use it. First we look at enterprise information in light of business goals in order to characterize the “unstructured” information gap. We then look at how that information, or at least the textual variety, may be structured for use. Then we look at a few uses, at enriching search, surely one of today’s killer apps, and at enhancing business intelligence via search.