Towards an Intelligent Information System Xiaofei Jia Mengchi Liu School of Computer Science Carleton University Ottawa, Ontario, Canada K1S 5B6 {xjia, mengchi}@scs.carleton.ca Abstract Nowadays, information overload is becoming an ever-growing problem. Finding the exact information is quite difficult. The main problem is that information is not organized according to its natural structures. There is not a data model that can represent these semi-structured or unstructured data in a natural way. On the other hand, almost every information application exploits interactive interfaces in some way. However, the development of user interfaces is still ad-hoc and often uses some predefined form-based interfaces. How can users express their queries in a semantic way is an important research topic. In this paper, we present an effective conceptual model to organize data according to their contents based on object-relation databases, and describe a quasi-natural language interface for the conceptual model. We also discuss the design and implementation of our prototype system. 1. Introduction Finding the exact information in this information overload era is usually a time consuming, tedious, and frustrating process. For example, people are getting used to searching the web as the first resort for information nowadays. However, few, if any, search engines consistently deliver the “right” information. What one could find from these search engines is usually over- whelming, but quite often irrelevant information. The main problem is that information is not organized according to its natural structures. There is not a data model that can represent these semi-structured or unstructured data in a natural way. Search engines use words in the docu- ments to index them, and use word matching and ranking techniques to generate search results. They only deal with words syntactically rather than semantically and thus cannot understand what users really want. This paper grows out of an interest in current web search and database research mainly object-relational databases (ORDB). Although ORDB have been considered as “the next great wave” [3] in the database community, the applications are difficult to develop as there is not a well-defined conceptual model for such kind of database design that plays the same role as ER or EER model [5] for the relational model. On the other hand, although database query languages, such as SQL, are very powerful for data access, real database users (end users) do not know how to use them. Almost every database application exploits some kind of interactive interfaces. However, the development of user interfaces is still ad-hoc [8] and often uses some predefined form-based style. This greatly limits what users can do with the data in databases. How can users express their queries in a semantic way is an important research topic. Natural language interfaces to databases (NLIDB) allow users to use natural language to express database queries so that users need not have the knowledge about the structures of the database. The successful development of NLIDB will greatly expand database applications.