Chinese/English Information Retrieval
PIRCS is an automatic, learning-based information retrieval system that is conceptualized as a three layer network and operates via activation spreading. It supports both ad-hoc retrieval and routing (filtering) operations. PIRCS was developed over the past years at Queens College, CUNY and has participated in all of the past annual TREC (Text REtrieval Conference) large scale blind retrieval experiments, and has consistently returned effectiveness among the best systems. PIRCS has also been extended with Chinese Language processing capability. In TREC-5 & 6 it participated in the first evaluated, large scale Chinese retrieval experiments with admirable results.
PIRCS is now being deveoped to support English-Chinese cross language retrieval capability for our paticipation in DARPA's TIDES project.
PIRCS stands for Probabilistic Indexing Retrieval - Components - System
Our objective is to investigate techniques to improve retrieval accuracy for both Chinese and English using PIRCS as our platform. Our current efforts include:
- Upgrade our research PIRCS system to handle collections of about 10 GB in size
- Upgrade our interactive PIRCS to a demonstrable English-Chinese cross language IR system
- Explore issues of question-answering in English and Chinese
- Adaptive filtering methodologies.