Published on October 3, 2007
Text Mining: Finding Nuggets in Mountains of Textual Data: Text Mining: Finding Nuggets in Mountains of Textual Data Jochen Dijrre, Peter Gerstl, Roland Seiffert Presented by Drew DeHaas Outline: Outline Motivation Methodology Feature Extraction Clustering and Categorizing Some Applications Conclusion & Exam Questions Motivation: Motivation A large portion of a company’s data is unstructured or semi-structured Letters Emails Phone recordings Contracts Technical documents Patents Web pages Articles Motivation: Motivation Rapid processing of large document collections Speed! Automation of tasks Objective analysis Typical Applications: Typical Applications Summarizing documents Discovering/monitoring relations among people, places, organizations, etc Organizing documents by content Indexing for search and retrieval Retrieving documents by content Outline: Outline Motivation Methodology Feature Extraction Clustering and Categorizing Some Applications Conclusion & Exam Questions Methodology: Challenges: Methodology: Challenges Information is in unstructured textual form Natural language interpretation is difficult & complex task! (not fully possible) Text mining deals with huge collections of documents Methodology: Two Aspects: Methodology: Two Aspects Knowledge Discovery Extraction of codified information Mining proper; determining some structure Information Distillation Analysis of feature distribution Two Text Mining Approaches: Two Text Mining Approaches Extraction Extraction of codified information from single document Analysis Analysis of the features to detect patterns, trends, etc, over whole collections of documents Comparison with Data Mining: Comparison with Data Mining Data mining Identify data set(s) Select features manually Prepare data Analyze distribution Text mining Identify documents Extract features Select features (automatically) Prepare data Analyze distribution IBM Intelligent Miner for Text: IBM Intelligent Miner for Text IBM introduced product in 1998 SDK with: Feature extraction, clustering, categorization, and more Traditional components (search engine, etc) No longer available? The rest of the paper describes text mining methodology of Intelligent Miner. Outline: Outline Motivation Methodology Feature Extraction Clustering and Categorizing Some Applications Conclusion & Exam Questions Feature Extraction: Feature Extraction Recognize and classify “significant” vocabulary items from the text Categories of vocabulary Proper names Multiword terms Abbreviations Relations Other useful things: numerical forms of numbers, percentages, money, etc Canonical Form Examples: Canonical Form Examples Normalize numbers, money Four = 4, five-hundred dollar = $500 Conversion of date to normal form Morphological variants Drive, drove, driven = drive Proper names and other forms Mr. Johnson, Bob Johnson, The author = Bob Johnson Feature Extraction Approach: Feature Extraction Approach Linguistically motivated heuristics Pattern matching Limited lexical information (part-of-speech) Avoid analyzing with too much depth Does not use too much lexical information No in-depth syntactic or semantic analysis Feature Extraction Example: Feature Extraction Example Disambiguating Proper Names (Nominator Program) Apply heuristics to strings, instead of interpreting semantics The unit of context for extraction is a document. The heuristics represent English naming conventions Advantages to IBM’s approach: Advantages to IBM’s approach Processing is very fast (helps when dealing with huge amounts of data) Heuristics work reasonably well Generally applicable to any domain Outline: Outline Motivation Methodology Feature Extraction Clustering and Categorizing Some Applications Conclusion & Exam Questions Clustering: Clustering Fully automatic process Documents are grouped according to similarity of their feature vectors Each cluster is labeled by a listing of the common terms/keywords Good for getting an overview of a document collection Two Clustering Engines: Two Clustering Engines Hierarchical clustering Orders the clusters into a tree reflecting various levels of similarity Binary relational clustering Flat clustering Relationships of different strengths between clusters, reflecting similarity Clustering Model: Clustering Model Categorization: Categorization Assigns documents to preexisting categories Classes of documents are defined by providing a set of sample documents. Training phase produces “categorization schema” Documents can be assigned to more than one category If confidence is low, document is set aside for human intervention Categorization Model: Categorization Model Outline: Outline Motivation Methodology Feature Extraction Clustering and Categorizing Some Applications Conclusion & Exam Questions Applications: Applications Customer Relationship Management application provided by IBM Intelligent Miner for Text called “Customer Relationship Intelligence” “Help companies better understand what their customers want and what they think about the company itself” Customer Intelligence Process: Customer Intelligence Process Take as input body of communications with customer Cluster the documents to identify issues Characterize the clusters to identify the conditions for problems Assign new messages to appropriate clusters Customer Intelligence Usage: Customer Intelligence Usage Knowledge Discovery Clustering used to create a structure that can be interpreted Information Distillation Refinement and extension of clustering results Interpreting the results Tuning of the clustering process Selecting meaningful clusters Outline: Outline Motivation Methodology Feature Extraction Clustering and Categorizing Some Applications Conclusion & Exam Questions Conclusion: Conclusion This paper introduced text mining and how it differs from data mining proper. Focused on the tasks of feature extraction and clustering/categorization Presented an overview of the tools/methods of IBM’s Intelligent Miner for Text Exam Question #1: Exam Question #1 Name an example of each of the two main classes of applications of text mining. Knowledge Discovery: Discovering a common customer complaint in a large collection of documents containing customer feedback. Information Distillation: Filtering future comments into pre-defined categories Exam Question #2: Exam Question #2 How does the procedure for text mining differ from the procedure for data mining? Adds feature extraction phase Infeasible for humans to select features manually The feature vectors are, in general, highly dimensional and sparse Exam Question #3: Exam Question #3 In the Nominator program of IBM’s Intelligent Miner for Text, an objective of the design is to enable rapid extraction of names from large amounts of text. How does this decision affect the ability of the program to interpret the semantics of text? Does not perform in-depth syntactic or semantic analysis of the text; the results are fast but only heuristic with regards to actual semantics of the text. Questions?: Questions?