Weimao Ke

Lixiao Yang, Mengyang Xu, Weimao Ke

Question-answering (QA) is an important application of Information Retrieval (IR) and language models, and the latest trend is toward pre-trained large neural networks with embedding parameters. Augmenting QA performances with these LLMs requires intensive computational resources for fine-tuning. We propose an innovative approach to improve QA task performances by integrating optimized vector retrievals and instruction methodologies. Based on retrieval augmentation, the process involves document embedding, vector retrieval, and context construction for optimal QA results. We experiment with different combinations of text segmentation techniques and similarity functions, and analyze their impacts on QA performances. Results show that the model with a small chunk size of 100 without any overlap of the chunks achieves the best result and outperforms the models based on semantic segmentation using sentences. We discuss related QA examples and offer insight into how model performances are improved within the two-stage framework.

Weimao Ke

We proposed a Least Information theory (LIT) to quantify meaning of information in probability distribution changes, from which a new information retrieval model was developed. We observed several important characteristics of the proposed theory and derived two quantities in the IR context for document representation. Given probability distributions in a collection as prior knowledge, LI Binary (LIB) quantifies least information due to the binary occurrence of a term in a document whereas LI Frequency (LIF) measures least information based on the probability of drawing a term from a bag of words. Three fusion methods were also developed to combine LIB and LIF quantities for term weighting and document ranking. Experiments on four benchmark TREC collections for ad hoc retrieval showed that LIT-based methods demonstrated very strong performances compared to classic TF*IDF and BM25, especially for verbose queries and hard search topics. The least information theory offers a new approach to measuring semantic quantities of information and provides valuable insight into the development of new IR models.