ICAR Research Data Repository for Knowledge Management
Optimizing Topic Coherence in the Gujarati Text Topic Modeling A Relevant Words Based Approach
Shodhganga@INFLIBNET
View Archive Info| Field | Value | |
| Title |
Optimizing Topic Coherence in the Gujarati Text Topic Modeling A Relevant Words Based Approach
— |
|
| Contributor |
Apurva Shah Gaurav Sharma
|
|
| Subject |
Topic Modeling, Latent Dirichlet Allocation,Text summerization, Dimensionality reduction
|
|
| Description |
quotTopic models have gained extensive consideration from the information retrieval research community. Topic models fundamentally transform high dimensional corpus to the low dimensional topic subspace. The low dimensional topic subspace is a set of a finite number of topics, which explains what the entire corpus is all about. The interpretability of the topic models indicates how good the whole corpus is being explained by the particular finite set of topics. It can be measured quantitatively by the semantic coherence of the topic model. newlineThe words of topics are placed in descending order according to the probability of words in that specific topic. The low probable words in the topic are less semantically relevant compared to the high probable words which result in the decrease of the semantic coherence. To improve the interpretability of the topic model, the semantic coherence optimization using relevant words technique has been proposed. newlineFurthermore, Gujarati linguistic knowledge has been incorporated for improving interpretability of the topic model. The method has been applied to reduce morphological inflectional forms of a word to its root word. The interpretability of the topic model is also influenced by poor quality topics such as identical topics or mixed topics. The techniques have been developed to eliminate poor quality topics. newlineThis PhD thesis would be useful for text analysis in various research domains where intensive text summarization as well as dimensionality reduction required. newlinequot newline newline References p. 93-102, Appendix_A p. 103-106 |
|
| Date |
2018-10-04T05:39:23Z
2018-10-04T05:39:23Z 02-05-2014 24/08/2018 — |
|
| Type |
Ph.D.
|
|
| Identifier |
http://hdl.handle.net/10603/218380
|
|
| Language |
English
|
|
| Relation |
No. of references 109
|
|
| Rights |
university
|
|
| Format |
xxv, 106p.
— None |
|
| Coverage |
Computer Engineering
|
|
| Publisher |
Ahmedabad
Gujarat Technological University Computer/IT Engineering |
|
| Source |
University
|
|