Abstract:

This project is focused on text summarization. Text summarization is the ability to obtain the key ideas from a text passage using as little words as possible. Dr. Pollett has a search engine, Yioop, that uses a centroid-based summarizer (CBS) to summarize its crawled documents. A CBS hinges on using a centroid (a set of words that are statistically important to the document) to get the main idea for the document. After that it computes the text frequencies and cosine similarity to build the summary. In CS297 I cloned the GIT Seek Quarry repository, performed a simple crawl and tested its summarizers using ROUGE. I also implemented a Dutch summarizer that is used in the Yioop search engine. Furthermore, I found a paper on a graph-based summarizer, coded it, tested it using ROUGE and added it to the Yioop search engine. Lastly, I implemented Dr. Pollett's summmarizer algorithm and added it to the Yioop search engine.

In this project I will continue to get deeper into text summarization. First, I will compare a few more summarizer methods. One of which is different than any that others that have been looked at so far. For example, a graph summarizer (which I already did), the centroid summarizer with weights and look into the famous summarizer Nick D'Aloisio wrote in the Summly app the he eventually sold to Yahoo for $30 million. In addition, I will do more extensive experiments. For example, using an existing set of summaries and documents like in the Document Understanding Conference (DUC) 2002 data set. Using a larger data set will allow us to really determine how good the summarizers we test perform.

CS297 Results

• Cloned the GIT Seek Quarry repository, performed a simple crawl and tested its summarizers using ROUGE
• Implemented a Dutch summarizer that is used in the Yioop search engine
• Found a paper on a graph-based summarizer, coded it, tested it using ROUGE and added it to the Yioop search engine
• Implemented Dr. Pollett's summmarizer algorithm and added it to the Yioop search engine

Proposed Schedule

Key Deliverables:

• Software
  • Dr. Pollett's algorithm completely implemented
  • Implementation of Nick D'Aloisio's summarizer integrated into Yioop
• Report
  • CS299 Report
  • Conduct and document all ROUGE experiment results
  • All code written during the semester and its documentation

Innovations and Challenges

• Getting Dr. Pollett's algorithm to produce better ROUGE results
• Implementing Nick D'Aloisio's summarizer and integrating it into Yioop
• Finding a large document set and writing the code to automate the testing the all of the summarizers against it

References:

[LinOch2004] Automatic Evaluation of Machine Translation Quality Using Longest Common Subsequence and Skip-Bigram Statistics. Chin-Yew Lin and Franz Josef Och. Association for Computational Linguistics. 2004.

[Lin2004] Looking for a Few Good Metrics: Automatic Summarization Evaluation How Many Samples Are Enough?. Chin-Yew Lin. NTCIR. 2004.

[Lovins1968] Development of a Stemming Algorithm. Julie Beth Lovins, Mechanical Translation and Computational linguistics. 1968.

[Porter1980] An algorithm for suffix stripping. M.F. Porter. Program: Electronic Library and Information Systems. 1980.

[Willett2006] The Porter stemming algorithm: then and now. Peter Willett. Program: Electronic Library and Information Systems. 2006.

[Porter2001] Snowball: A language for stemming algorithms. M.F. Porter. Snowball. 2001.

[Samei2014] Multi-Document Summarization Using Graph-Based Iterative Ranking Algorithms and Information Theoretical Distortion Measures. Borhan Samei, Marzieh Eshtiagh, Fazel Keshtkar and Sattar Hashemi, The Florida AI Research Society. 2014.

[Agrawal2014] A Graph Based Ranking Strategy for Automated Text Summarization. Nitin Agrawal, Shikhar Sharma, Prashant Sinha, Shobha Bagai, DU Journal of Undergraduate Research and Innovation. 2014.

[Langvielle2006] Google's PageRank and Beyond: The Science of Search Engine Rankings. Amy Langville and Carl Meyer.Princeton University Press.2006.

[Cutler1997] Using the Structure of HTML Documents to Improve Retrieval. Michal Cutler, Yungming Shih and Weiyi Meng. Proceedings of the USENIX Symposium on Internet Technologies and Systems. 1997.

[Bassil 2012] Semantic-Sensitive Web Information Retrieval Model for HTML Documents. Youssef Bassil and Paul Semaan. European Journal of Scientific Research. 2012.

[Bonnington 2011] Teen's IOS App Uses Complex Algorithms to Summarize the Web. Christina Bonnington. Wired.com. 2011.