Return to main listing.
Svetlozar Nestorov , Computation Institute, U of Chicago
Date Posted: November 3, 2009Application Deadline: March 31, 2010
Contact Information: Svetlozar Nestorov evtimov@cs.uchicago.edu
Description:
Svetlozar Nestorov from the Computation Institute is looking for students interested in indepedent study or paid internship opportunities for the Expinions project. Two possible projects are list below.If you are interested in these opportunities, please contact Mr. Nestorov (evtimov@cs.uchicago.edu).
Project 1: Adaptive query rewriting and expansion for online search APIs
The Expinions (expert opinions) project seeks to discover, extract and organize expert opinions from the vast amount of online documents including presentations, papers, blogs, and discussion forums.
The initial step is the discovery of relevant urls by querying search engines such as bing and google. The search APIs of these search engines have varying degrees of flexibility and often return a limited number of results for any given query.
The goal of the proposed student project is to develop and implement a system for rewriting and expanding queries to circumvent some of the limitations imposed by online search APIs.
For example, Google Search API returns only the first 64 results of any query but allows unlimited number of different queries. Starting with a base query such as "autism hypothesis", one can use the first 64 results to generate additional relevant terms and generate a number of new expanded queries. Sending these new queries to Google will likely produce new results which in turn can be used to generate more expanded queries.
Project 2: Online document annotation interface
The Expinions (expert opinions) project seeks to discover, extract and organize expert opinions from the vast amount of online documents including presentations, papers, blogs, and discussion forums.
In order to develop machine learning algorithms that automatically discover relevant statements in documents, we need to generate a large number of training examples. The training examples will be created by individuals using an online document annotation interface.
The goal of the proposed project is to develop javascript code that enables users to annotate text web pages. The web pages will be displayed by the CI web server and will be instrumented with the developed javascript code. Users should be able to cross out paragraphs and sentences they consider irrelevant as well as select parts of the text they term relevant. Users should also be able to modify their previous selections. The javascript code should change the appearance of the text in the web browser in response to user actions. It should also interface php services in order to record and save users work.
If this job has been filled or you find a broken link, please contact Sharon Salveter to get the page updated.