Sunday, 28 June 2015

Document Clustering Literature Survey

Introduction


Text Mining is the discovery of unknown information, by automatically extracting information from different written resources. Text mining is different from what are familiar with in web search. Text mining is a variation on a field called data mining that tries to find interesting patterns from large databases. Text mining is also known as Intelligent Text Analysis, Text Data Mining or Knowledge-Discovery in Text (KDT). Text mining is a young interdisciplinary field which draws on information retrieval, data mining, machine learning, statistics and computational linguistics.
Document Clustering is the application of cluster analysis to textual documents. It has application in automatic document organization, topic extraction and fast information retrieval or filtering. In the world of computer technology every 10 minutes billions of gigabytes of data is generated and people must be able to retrieve this data efficiently. It generates the need of the document clustering i.e. also the systematic organization of data. Many Search Engines like Google, Yahoo, Baidu, Bing, AltaVista and even many commercial search engines are available to organize the data into useful format. Even at Intranet, clustering of data is becoming highly important. Besides the different nature of Clustering on Intranet and Internet, Basic Requirement is almost the same.
The aim of this report is to specify the entire existing algorithms for document clustering. Document clustering algorithms have evolved over time gradually into a more subtle way. The quality of Document clustering is defined by the fulfillment of user needs.  Initially queries were matched in Boolean form. After that Relevance Feedback came into existence. Further Search Strategies were improved using Latent Semantic Indexing and Non-negative Matrix Factorization. The things even got totally changed with the start of machine learning algorithms. The new field of Natural Language Processing (NLP) came into existence for the organized study of Text Mining. The algorithm like K-means and fuzzy c- means were used now to find the nearest neighbors for Document Clustering. All the Algorithms used And Categorization of Text Mining has been specified in this document.
The Scope of this project in Searching and querying, Ranking of search results, Navigating and browsing information, Optimizing information representation and storage, Document classification (into predefined group), Document clustering (automatic discovered results ) .



       I.            Literature Survey

Boolean logic

For document clustering technique initially Boolean logic was used for comparison purpose. It is used now also but not exactly in the same form. For the clustering of document only  Boolean comparison was not sufficient. Since always data is not available in the structured format. Also Document clustering result is not meant only for the query given. But it should be in the form what user actually needs.

Relevance Feedback

Relevance feedback is a feature of some information retrieval systems. The idea behind relevance feedback is to take the results that are initially returned from a given query and to use information about whether or not those results are relevant to perform a new query.

Latent Semantic Indexing

Latent semantic indexing (LSI) is an indexing and retrieval method that uses a mathematical technique called singular value decomposition (SVD) to identify patterns in the relationships between the terms and concepts contained in an unstructured collection of text. A key feature of LSI is its ability to extract the conceptual content of a body of text by establishing associations between those terms that occur in similar contexts.LSI overcomes two of the most problematic constraints of Boolean keyword queries: multiple words that have similar meanings (synonymy) and words that have more than one meaning (polysemy). Synonymy is often the cause of mismatches in the vocabulary used by the authors of documents and the users of information retrieval systems. As a result, Boolean or keyword queries often return irrelevant results and miss information that is relevant.

Non-Negative Matrix Factorization

While using Latent Semantic Indexing Every word does not have the equal importance to the user queries. So they must be given different weightage on the basis of term frequency which must be taken from the document. It is done by using Non-Negative Matrix Factorization.
In this process document term matrix is constructed with the weights of various terms from a set of documents. This matrix is factored into a term-feature and a feature-document matrix. The features are derived from the contents of the documents, and the feature-document matrix describes data clusters of related documents.

   II.            Technological Foundation of Machine Learning

The field of natural language processing has produced technologies that teach computers natural language so that they may analyze, understand, and even generate text. Some of the technologies that have been developed and can be used in the text mining process are information extraction, topic tracking, summarization, categorization, clustering, concept linkage, information visualization and question answering.

Information Extraction

Information extraction identifies key phrases and relationships within text. It is done by looking for predefined sequences in text, a process called pattern matching. For information Extraction, Rule Based mining algorithms For ex. Sequential Covering  are used.
     

Topic Tracking

A topic tracking system works by keeping user profiles and, based on the documents the user views ,predicts other documents of interest to the user. Google offers a free topic tracking tool that allows users to choose keywords and notifies them when news relating to those topics becomes available.
Term Frequency –Inverse Document Frequency (TF IDF) algorithm is used for topic tracking. This algorithm is numerical statistics that reflects how important a word is to document. We used weighting factor for each term on the basis of domain knowledge and term frequency in document.

Summarization

The key to summarization is to reduce the length and detail of a document while retaining its main points and overall meaning. The challenge is that, although computers are able to identify people, places, and time, it is still difficult to teach software to analyze semantics and to interpret meaning. It uses fuzzy c- mean algorithm.

Categorization

Categorization involves identifying the main themes of a document by placing the document into a pre-defined set of topics. Categorization often relies on a thesaurus for which topics are predefined, and relationships are identified by looking for broad terms, narrower terms, synonyms, and related terms. Categorization tools normally have a method for ranking the documents in order of which documents have the most content on a particular topic. It uses Support Vector machine.

Clustering

Clustering is a technique used to group similar documents, but it differs from categorization in that documents are clustered on the fly instead of through the use of predefined topics. A basic clustering algorithm creates a vector of topics for each document and measures the weights of how well the document fits into each cluster.
(1) K-means clustering algorithm
(2) Word relativity-based clustering (WRBC) method, text clustering process contains four main parts: Text reprocessing, word relativity computation, word clustering and text classification. Remove stop-words, Stemming, Filtering

Concept Linkage

Concept linkage tools connect related documents by identifying their commonly shared concepts and help users find information that they perhaps wouldn’t have found using traditional searching methods. Concept linkage is a valuable concept in text mining, especially in the biomedical fields where so much research has been done that it is impossible for researchers to read all the material and make associations to other research. Ideally, concept linking software can identify links between diseases and treatments when humans cannot.

Information Visualization

Visual text mining, or information visualization, puts large textual sources in a visual hierarchy or map and provides browsing capabilities, in addition to simple searching. The  information visualization may be conducted into three steps: (1) Data preparation: i.e. determine and acquire original data of visualization and form original data space. (2) Data analysis and extraction: i.e. analyze and extract visualization data needed from original data and form visualization data space. (3) Visualization mapping: i.e. employ certain mapping algorithm to map visualization data space to visualization target.  

Question Answering

Another application area of natural language processing is natural language queries, or question answering (Q&A), which deals with how to find the best answer to a given question. Many websites that are equipped with question answering technology, allow end users to “ask” the computer a question and be given an answer. Q&A can utilize multiple text mining techniques.

III.            The Evolution of Document Clustering

Document Clustering is the application of cluster analysis to textual documents. Document Clustering involves the use of Descriptors and Descriptor Extraction. Descriptors are set of words that describe the contents within the clusters. Application of document clustering is done in the following fields.
1.      Automatic Document Organization
2.      Topic Extraction
3.      Fast Information Retrieval
4.      Filtering

Algorithms Used for Document Clustering are:

1.      Hierarchical Based Algorithms          
2.      K-Means Algorithms and it’s Variants
3.      Other Algorithms are:
a)      Graph Based Algorithms
b)      Ontology Supported Algorithms
c)      Order Sensitive Clustering

Hierarchical Clustering Algorithms (HCA)

In data mining Hierarchical Clustering Algorithms is a method of cluster Analysis which seeks to build hierarchy of Clusters. Two Ways of Hierarchical Clustering Algorithms are:
1.      Agglomerative: This is a "bottom up" approach. each observation starts in its own cluster, and pairs of clusters are merged as one moves up the hierarchy.
2.      Divisive: This is a "top down" approach: all observations start in one cluster, and splits are performed recursively as one moves down the hierarchy.

K-Means And it’s Variants

k-means clustering is a method of vector quantization, originally from signal processing, that is popular for cluster analysis in data mining. k-means clustering aims to partition n observations into k clusters in which each observation belongs to the cluster with the nearest mean, serving as a prototype of the cluster. This results in a partitioning of the data space into Voronoi cells
1.      Fuzzy C-Means Clustering is a soft version of K-means, where each data point has a fuzzy degree of belonging to each cluster.

2.      The filtering algorithm uses kd-trees to speed up each k-means step.

No comments:

Post a Comment