Meeting
AgendaMeeting
Agenda
Now would be a very good time to let me know if there is software you would like to see on CGL machines. When I leave, Wilkin will take over the critical aspects of the job, but he might not have time for very many non-mandatory software updates.
Thursday, August 22, 1996 Symbolic Computation Seminar "The implementation for MAPLE of the new diffalg package" Francois Boulier, Symbolic Computation Group Computer Science Dept., University of Waterloo 10:30 a.m.; DC 1304 MASTER'S THESIS PRESENTATION Elizabeth Tudhope, graduate student, Dept. Comp. Sci., Univ. Waterloo, will speak on ``Query Based Stemming''. DATE: Tuesday, August 27, 1996 TIME: 1:00-2:00 p.m. ROOM: DC 1304 ABSTRACT In information retrieval (IR) the decision of whether a document is relevant for a given query is generally determined by the number and frequency of terms the document and query have in common. Query expansion is a method of increasing the number of documents matched by a query. The addition of well selected terms to the original query provides the opportunity for more relevant documents to match. One method of query expansion is the use of stemming to add morphological variants of query terms. Stemming is a process used to reduce word forms to common roots. In IR systems stemming is usually performed at the document indexing phase. The terms in the document are stemmed and the index is built based on the resulting stems. Stemming at index time fixes an equivalence relationship between all words that share a common root. In this talk we will present an overview of stemming as a query expansion process and discuss some of the limitations/problems of current stemming techniques. Next an architecture for a system that performs stemming at query time will be presented. Finally, we will discuss the effectiveness of applying stemming at query time to avoid some of the the inherent problems of index time stemming. ------------------------------------------------------- MASTER'S THESIS PRESENTATION Shirley Yuk Pik Shek, graduate student, Dept. Comp. Sci., Univ. Watelroo, will speak on ``TAR-tree: A Spatial Indexing Method''. DATE: Wednesday, August 28, 1996 TIME: 11:00 a.m. - 12:00 noon ROOM: DC 1331 ABSTRACT TAR-tree, which stands for local minimal area Triangle and Aligned Rectangle tree, is a new dynamic access method for spatial objects. It is a height-balanced tree whose leaves contain spatial objects and whose interior nodes correspond to hierarchies of nested triangles and aligned rectangles. Both point and region queries are allowed. Triangular bounding boxes can often provide a greater selectivity in the search. In this presentation, I will give the algorithms for insert, search and delete operations. The experimental results of storage requirement, construction time and search performance comparisons of TAR-tree with quadratic R-tree and R*-tree are also presented. Experiments were carried out on real data obtained from Faculty of Environmental Studies. Real data files form part of Waterloo County. ------------------------------------------------------- Master's Thesis Presentation Biraj Saha, Graduate Student, Department of Computer Science, University of Waterloo will speak on ``A Mapping of Object Schema to Existential Graphs''. DATE: Wednesday, August 28, 1996 TIME: 2:00 - 3:00 p.m. ROOM: Davis Centre Room DC3301 ABSTRACT This thesis presents a formalism called ``UDR existential graphs''(UEG) based on the Beta extension to the existential graph formalism developed by Peirce. The purpose is to use this formalism as a basis for modeling object schema such that an external application can reason about its declarations, particularly those being ``constraints.'' In my presentation, I shall outline the formalism and show how to map a given schema to a UEG in such a way that external applications can reason about this representation for purposes such as semantic query optimization, schema evaluation, etc. I shall also outline a method for "graphical unit resolution inference" in a UEG that serves as a key in supporting incremental update to the representation of a given schema. -------------------------------------------------------