DataparkSearch at TREC 2005

QuALiM at TREC 2005: Web-Question Answering with FrameNet

In this paper I describe my TREC 2005 participation. The system used was–except from one new module–the same as in TREC 2004. In the following I will describe this new module, which uses the annotated Natural Language data collected in the FrameNet project in order to find paraphrases to answer questions. I will furthermore present and discuss the TREC 2005 results and compare them to those ach...

QACTIS-based Question Answering at TREC 2005

The QACTIS system is being developed for the eventual purpose of providing a user the capability of multilingual question-answering from multimedia. QACTIS was tested at TREC-2005 as a means of identifying its successes and limitations in answering questions specifically from English newswire text as it moves in the direction of multilingual , multimedia question answering. In this paper, we pr...

CSUSM at TREC 2005: Genomics and Enterprise Track

MG4J at TREC 2005

MG4J participated in two tracks of TREC 2005 — the ad hoc task and the efficiency task of the Terabyte Track (find all the relevant documents with high precision from 25.2 million pages from the .gov domain). It was the first time the MG4J group participated to TREC, and we concentrated our efforts on the ad hoc task, using a combination of techniques based on a new multi-index minimal-interval...

JAVELIN I and II Systems at TREC 2005

The JAVELIN team at Carnegie Mellon University submitted three question-answering runs for the TREC 2005 evaluation. The JAVELIN I system was used to generate a single submission to the main track, and the JAVELIN II system was used to generate two submissions to the relationship track. In the sections that follow, we separately describe each system and the submission(s) it produced, and conclu...