ch7 (1)

Creating Value With Social Media Text Analytics

• Author: Dr. Gohar F. Khan

• Chapter: 7

Learning Outcomes

Upon concluding this chapter, readers will have gained the knowledge and skills to:

• Comprehend basic social text analytics concepts and tools.

• Understand uses of text analytics by different sectors:

  • Business

  • Government

  • Academia

  • Financial institutes

• Recognize objectives of social media text analytics for business intelligence, including:

  • Sentiment analysis

  • Concept mining

  • Trends mining

  • Topic mining

• Comprehend the text analytics cycle and steps required to extract business insights from text.

• Understand various text analytics terms, methods, and algorithms.

• Extract and analyze social media text.

• List common limitations and issues in text analytics.

Unstructured Data Importance

• Fact: 80% of data is unstructured.

  • Examples include:

    • Database notes

    • Call center transcripts

    • Emails

    • Open-ended survey responses

    • Web pages

    • News groups

    • Reviews, tweets, comments

    • Multimedia (photos, videos, infographics)

• Consequence: Decision-makers rely on only 20% of data.

Introduction to Social Media Text Analytics

• Text Definition: Fundamental element of social media platforms includes comments, tweets, blog posts, product reviews, and status updates.

• Social Media Text Analytics (Text Mining): Technique to extract, analyze, and interpret hidden business insights from textual data.

Purpose of Text Analytics

• Categories of Uses:

  • By Business

  • By Government

  • By Academia

  • By Financial Institutes

Types of Social Media Text

Dynamic Text

• Definition: Real-time, user-generated content articulating opinions on various topics.

• Characteristics:

  • Shorter in length (e.g., a couple of sentences)

  • Frequently updated or deleted.

• Examples:

  • Tweets

  • Comments

  • Discussions

  • Reviews

Static Text

• Definition: Text that is revised infrequently.

• Characteristics: Longer format (e.g., several paragraphs).

• Examples:

  • Wiki content

  • Blogs

  • Word documents

  • Corporate reports

  • News transcripts

Text Analytics in the Industry

• IBM: Leading provider with enterprise-level text analytics solutions via NLP tools.

• Google: Offers Cloud Natural Language API for understanding customer conversations and documents.

• Microsoft: Azure Text Analytics provides services for sentiment analysis, key phrase extraction, etc.

• SAS: Advanced analytics platform for discovering patterns in text data.

• Smaller Specialists: Aylien, TextRazor focusing on tailored NLP services.

Deployment Models

• On-premise Model: Expensive but offers greater security and control.

• Cloud-based Model: Cost-effective, scalable, and lower risks; attractive for small/medium businesses.

Applications for Text Analysis

• Common Applications:

  • Document management

  • Corporate history

  • Scientific publications

  • Thematic understanding of websites

  • Survey data

  • Email comprehension

  • Call center data

Text Analytics Purposes

Sentiment Analysis

• Definition: Assess social media text (mostly dynamic) as positive, negative, or neutral.

Intention Mining

• Purpose: Discovers user intentions (e.g., buying, selling, recommending) from media text.

Trends Mining

• Definition: Predictive analytics to foresee future events using vast amounts of social media data.

Concept Mining

• Purpose: Extracts ideas/concepts from static social media text for classification and clustering.

Text Analytics Mechanism

Phases of Text Analytics Cycle

1. Identification and Searching: Locate the text source for analysis, acknowledging diversity and noise in social media content.

2. Text Parsing and Filtering: Clean, filter, and prepare text using NLP techniques to remove irrelevant elements.

3. Text Transformation: Convert cleaned text into a computer-readable format (binary code).

4. Text Mining: Apply various algorithms (clustering, association, classification, prediction) to extract business insights.

Common Text Analytics Terms

• Natural Language Processing (NLP)

• Information Retrieval (IR)

• Named Entity Recognition (NER)

• Corpus

• Bag of Words (BoW)

• Latent Semantic Analysis (LSA)

• Latent Dirichlet Allocation (LDA)

Challenges in Text Analytics

• Lack of a solid business case

• Resource intensity

• Complexity of sentiments

• Contextual nature of data

• Issues with multilingual text

• Cultural/regional differences

Text Analytics Tools

• NLTK: Python library for NLP tasks.

• spaCy: Production-ready Python NLP library with pre-trained models.

• Gensim: Focused on topic modeling and document similarity.

• TextBlob: Simple library for various NLP tasks including sentiment analysis.

Social Media Text Analysis Tools

• Lexalytics: Tool for semantic analysis and sentiment extraction.

• Discovertext: Platform for collecting and analyzing text streams.

• Tweet Archivist: Tool for archiving and analyzing tweets.

• Twitonomy: Detailed analytics of Twitter engagements.

Review Questions

• Discuss the usefulness of text analytics and provide a differentiation between static and dynamic text.

• Explain the four primary purposes of social media text analytics.

• Distinguish between supervised and unsupervised machine learning techniques.

• Describe the social media text value creation cycle.

• Identify issues related to text analytics.