Suggesting Topic-Based Query Terms as You Type

摘要: Query term suggestion that interactively expands the queries is an indispensable technique to help users formulate high-quality and has attracted much attention in community of web search. Existing methods usually suggest terms based on statistics documents as well query logs external dictionaries, they neglect fact topic information very crucial because it helps retrieve topically relevant documents. To give gratification, we propose a novel method: user types letter by letter, are coherent with could instantly. For effectively suggesting highly terms, generative model incorporating topical coherence terms. The learns topics from underlying Latent Dirichlet Allocation (LDA). achieving goal instant suggestion, use trie structure index access We devise efficient top-k algorithm type queries. Experimental results show our approach not only improves effectiveness but also achieves better efficiency scalability.