IR Yi Zhang's term project
Basic Information
- Project Title: A
Generative Model for Generic Text Summarization
- Name: Yi Zhang (yiz@cs.cmu.edu)
(cooperate with Jade Goldstein)
- Presentation Date: Dec. 7 or Dec. 12
Contents
Abstract
For
most of the text summarization task, people approach the problem by generating
summaries from the document, which is usually the way human abstracters do using
a model like:
Document ->
Abstract
On the other hand, when writing an article, such as a research paper, people often first have some idea about what is the main idea, then write an outline, finally write the whole article according to the outline. we can view the process as a generative model like:
Semantic
structures -> Outline (Abstract
)-> Document (
)
In my project, I will use this model to do text summarization task
Proposal
Timelines
Task |
to be done by |
status |
| Set up the web page | Sept. 27 | done |
| Reading relevant papers | Sept. 27 | done |
| Corpus | Oct. 3 | done |
| Finding useful tools | Oct. 31 | done |
| System Design | Oct. 31 | done |
| Coding | Nov. 20 | done |
| Result analysis | Dec. 1 | done |
| Write-up, presentation | Decl 7th | done |
System Description
Part1: Neural Network
Part2: Maximum Entropy Model
Part3: Combination of NN result and ME result
Presentation and Results
Final report
Conclusions
Reference
A maximum entropy approach to natural language processing , by Adam L. Berger
OCELOT: A system for summarizing web pages by Adam L.Berger, Vibhu O. Mittal
A Comparision of Ranking Rankings Produced by Summarization Evaluation Measures by Robert L. Donaway ,Kevin W. Drummey, Laura A. Mather
Summarizing Text
Documents: Sentence Selection and Evaluation Metrics , by Jade
Goldstein, Mark Kantrowitz, Vibhu Mittal and Jaime Carbonell , SIGIR 99
The Use of MMR,Diversity-based Reranking for Reording Documents and Producing Summaries,
by Jade Goldstein , SIGIR 98