Successfully reported this slideshow.
Your SlideShare is downloading. ×

Text semantics with azure text analytics cognitive services

Ad

NLP : TEXT ANALYTICS WITH
AZURE COGNTIVE SERVICES
PRIYANKA H SHAH

Ad

I am…
 Head of Innovation and Technology.
 AI/ML solution architect and Big data solution engineer.
 Full stack develop...

Ad

THE DATA
EXPLOSION
PROBLEM
Huge data available:
Text documents
Anomly / Error logs
Earnings call transcripts
Annual Report...

Ad

Ad

Ad

Ad

Ad

Ad

Ad

Ad

Loading in …3
×

Check these out next

1 of 11 Ad
1 of 11 Ad

Text semantics with azure text analytics cognitive services

Download to read offline

Text analytics, sentiment detection etc. form a very important part of NLP. With huge data available today, making sense of the dara, detecting latent patterns, tagging documents etc. has become an important part of language modelling. Let us see how Azure Cognitive services enable us to detect document sentiments, Named entities, Key phrases, contextual information like geo spatial indices etc.

Text analytics, sentiment detection etc. form a very important part of NLP. With huge data available today, making sense of the dara, detecting latent patterns, tagging documents etc. has become an important part of language modelling. Let us see how Azure Cognitive services enable us to detect document sentiments, Named entities, Key phrases, contextual information like geo spatial indices etc.

Advertisement
Advertisement

More Related Content

More from CodeOps Technologies LLP

Advertisement

Text semantics with azure text analytics cognitive services

  1. 1. NLP : TEXT ANALYTICS WITH AZURE COGNTIVE SERVICES PRIYANKA H SHAH
  2. 2. I am…  Head of Innovation and Technology.  AI/ML solution architect and Big data solution engineer.  Full stack developer.  Passionate about new technologies, love to code, blog / talk about AI, ML.NET, Microsoft technology stack.  Environment enthusiast  Twitter: @fuzzymind1
  3. 3. THE DATA EXPLOSION PROBLEM Huge data available: Text documents Anomly / Error logs Earnings call transcripts Annual Reports Potential of data is untapped Latent patterns unexplored Quick ways to leverage the data potential?
  4. 4. REAL TIME BUSINESS CASE Built a document text search with Elastic Search Search efficiency for term search, phrase search, fuzzy search was quite good • Driverless cars -> Self driving cars, robotic cars • Lithium -> Other metals/ Utilities sector related information, dry cell or related industries Related terms search? • like were you looking for <term> OR • People who searched this report, also searched <list of related reports> Recommend search terms for sematic searches
  5. 5. Other constraints 550,000 call reports available Distributed across 11 sectors New call reports added daily Impossible to go through each document and identify the “tag” words How to tag newly uploaded call reports? How to establish word semantic relationships / build a word taxonomy for all documents
  6. 6. AZURE TEXT ANALYTICS COGNITIVE SERVICE •sentiment analysis, •opinion mining, •key phrase extraction, •language detection, •named entity recognition. cloud-based service that provides Natural Language Processing (NLP) features for text mining and text analysis
  7. 7. FEATURES OF TEXT ANALYTICS IDENTIFY AND CATEGORIZE IMPORTANT CONCEPTS BETTER UNDERSTAND CUSTOMER PERCEPTION DETECT LANGUAGE OF YOUR TEXT EXTRACT KEY PHRASES IN UNSTRUCTURED TEXT PROCESS UNSTRUCTURED MEDICAL DATA DEPLOY ANYWHERE, FROM THE CLOUD TO THE EDGE
  8. 8. DEMO
  9. 9. COMPLETE SOLUTION ARCHITECTURE Identify topics using LDA Build word taxonomy using the topics and the words occurring in each topic For example : Insurance, premium, policy, claim, reinsurance can be added as synonyms Use Azure Cognitive text Analytics services to get: Sentiment for each document Named Entity recognition Key phrases Contextual information about NERs Use the text Analytics information to tag documents Feed the tags and synonyms to Azure Cognitive search using Synonyms API Iteratively build a more sophisticated taxonomy. Keep refining every month for new reports
  10. 10. DETAILED SESSION LINK: • https://www.youtube.com/watch?v=S MfN4eSS0PA
  11. 11. THANK YOU

×