Cluster-based language models for distributed retrieval
Proceedings of the 22nd annual international ACM SIGIR conference on Research and development in information retrieval
| Hi-index | 0.25 |
This article presents a system designed to substitute current centralized Internet search engines with a distributed indexing scheme in a gradual sequence of steps. The advantage of this transition process is that each of the steps is relatively simple and at the same time makes a significant contribution to the quality of the whole system, resulting in a system with remarkable properties. It is scalable and automatically adaptive, and it facilitates not only the search for information, but also cooperative information filtering and decentralized publishing. Current technology is based on the use of large centralized databases and has potential scalability problems which will become apparent with the further growth of the Internet. This is why the authors believe that sooner or later the transition process presented in this article will take place. The main principles of the new distributed system are presented and discussed. The authors also identify the problems environments such as the Internet present for resource discovery systems. A number of resource discovery projects targeting these problems are analyzed in the second part of the article, and their strengths and disadvantages are identified