State-of-the-art statistical approaches to the Coreference Resolution task rely on sophisticated modeling, but very few (10-20) simple features. In this paper we propose to extend the standard feature set substantially, incorporating more linguistic knowledge. To investigate the usability of linguistically motivated features, we evaluate our system for a variety of machine learners on the stand...

CLaC-CORE, an exhaustive feature combination system ranked 4th among 34 teams in the Semantic Textual Similarity shared task STS 2013. Using a core set of 11 lexical features of the most basic kind, it uses a support vector regressor which uses a combination of these lexical features to train a model for predicting similarity between sentences in a two phase method, which in turn uses all combi...

This year, we participated in TRECVID 2010 content-based copy detection task. In this notebook paper we will describe our work in details. Different from last year when we just used SURF (speeded up robust feature) as visual feature, this year we employed a combination of four different features for our rough detection process: global SURF, center SURR, global color correlogram and center corre...

The culmination of the 2002 RoboFlag Summer Undergraduate Research Fellowship program, jointly operated between California Institute of Technology and Cornell University, was a final competition between two teams of three undergraduate researchers. After ten weeks of preparation, Team Pasadena defeated Team Ithaca in two of the three final games. This paper provides the detailed results of the ...

The relationship between crab size and shell size selected, in a free-choice situation, was tested for crabs maintained under several conditions . Individuals of Calcinus tibicen that were held and presented with an excess of shells for 7 days prior to testing selected smaller shells than control crabs (either those maintained for 7 days with no excess shells available or those experiencing nat...