21st February 2002

Due Tuesday, April 16 2002

* Projects carry 20% of the course credit

* Projects are due on April 16 positively

* You should give a demo before the project due date

• The idea behind the class project is to allow you to explore, in some depth, the topics covered in the class.
• Abstract due on 12th March: Mail me a one page write-up clearly defining your problem. Outline the approach you will follow in solving the problem and your plan of work.
• Literature Survey: You should read 3-4 key papers on your topic to understand the current state of the art. Then develop a system that implements a solution for the problem.
• Programming: All code that you write on your own should be well documented. Feel free to use code from the assignments. You can also use code from other sources but properly acknowledge the source.
• Tools and Data: Clearly state the sources of your data and also any tools you use in implementing your system. Identify your training and testing data clearly.
• Project Report: Write a project report on the literature survey, your methodology wherein you should discuss your algorithm and its implementation, results and conclusions. Also include any assumptions you have made. In your conclusions clearly state your achievements and also the limitations of your system. Also list any efforts that resulted in failure. There is no limit on the report but a well written report of about 5 pages is what I am expecting.
• One Page Summary: Include: Title, the exact problem that you have solved, description of your algorithm, the training data, the testing data, results. Also any other information you consider important. This is in addition to the project report.
• Submissions: Mail me a well written report, summary and well documented code.
• Demo: Give a demo of the system you have developed. Plan for a demo before the submission date.
• Grading: Project constitutes 20% of the course marks. Of this, your understanding of prior art on the topic will constitute 50% and the other 50% will be brought up by the actual implementation.
• Project Topics:
  1. Human Computer Interaction: Design and implement a command and control system using Latent Semantic Analysis for interacting with your computer.
  2. Text Summarization: Implement a text summarization system that takes documents and summarizes them down to a few sentences.
  3. Incorporating linguistic knowledge into n-gram Language Models:
  4. Grammar and Spelling Checker: Implement a system to detect spelling and syntax errors and correct them.
  5. Text Categorization: Implement an application of text categorization for the detection of spam (unsolicited bulk email).
  6. Text Tiling/Segmentation: Implement a multiparagraph segmentation algorithm that detects topic changes between paragraphs.
  7. Machine Translation: Implement a statistical machine translation system.
  8. Information Retrieval: Develop a search engine using Latent Semantic Indexing for the CSE department.
  9. Information Extraction: Design a system to look at conference announcements available on the net to extract the following information: Conference title, conference dates, submission deadline, venue, etc.
  10. Modifications of above projects or new ones that you can think of. Please discuss with me before you go ahead.