The HLTCOE Approach to the TREC 2012 KBA Track
نویسندگان
چکیده
Our team submitted runs for the TREC KBA Cumulative Citation Recommendation task. This task involves labeling over 300 million documents for whether they are relevant and/or central to particular entities already in a database. For this task, we used an SVM classifier that uses unigrams and named entities as binary features. In this paper, we describe our work for the 2012 evaluation and the results we obtained.
منابع مشابه
CWI at TREC 2012, KBA Track and Session Track
We participated in two tracks: Knowledge Base Acceleration (KBA) Track and Session Track. In the KBA track, we focused on experimenting with different approaches as it is the first time the track is launched. We experimented with supervised and unsupervised retrieval models. Our supervised approach models include language models and a string-learning system. Our unsupervised approaches include ...
متن کاملA Related Entity based Approach for Knowledge Base Acceleration
In this paper we present the overview of our work in the TREC 2013 KBA Track. The goal is to find documents which may contribute to the update of knowledge base entries (e.g., Wikipedia or Freebase articles). Two tasks are introduced in this year’s track: (1) Cumulative Citation Recommendation (CCR), (2) Streaming Slot Filling (SSF). Particularly, we focus on the CCR task, follow our previous w...
متن کاملPRIS at TREC 2012 KBA Track
Our system to KBA Track at TREC2012 is described in this paper, which includes preprocessing, index building, relevance feedback and similarity calculation. In particular, the Jaccard coefficient was applied to calculate the similarities between documents. We also show the evaluation results for our team and the comparison with the best and median evaluations.
متن کاملA Pattern Matching Approach to Streaming Slot Filling
In this paper, we described our system for Knowledge Base Acceleration (KBA) Track at TREC 2013. The KBA Track has two tasks, CCR and SSF. Our approach consists of two major steps: selecting documents and extracting slot values. Selecting documents is to look for and save the documents that mention the entities of interest. The second step involves with generating seed patterns to extract the s...
متن کاملK2U at TREC 2014 KBA Track
There are two types of nodes, called “spouts” and “bolts”. A spout is a source of streams (sequences of tuples). In case of the KBA track, a spout would read document data from the provided KBA corpus and emit them as a stream. A bolt receives any number of input streams, does some processing, and may emit new streams. For the KBA track, bolts would determine whether inbound documents from the ...
متن کامل