Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.

Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.

Similarity queries in linear constraint databases

Ying Deng, University of Nebraska - Lincoln

Abstract

The constraint database model has recently been proposed as a promising framework to model continuous variables and therefore images and spaces. Constraint databases are being increasingly used to solve spatial problems. This dissertation presents further advances on the use of the constraint data model in the context of similarity retrieval. Similarity-based retrieval is an important task in many spatial and image database applications. Given a representation scheme, it is necessary to employ some measure to determine the visual or spatial similarity of the objects. Spatial similarity is complex due to the numerous constraining properties of spatial objects and their embedding in space. We propose a new method to measure the similarities between two 2-dimensional spatial scenes that are composed of spatial objects, which are a set of points, line segments or polygons. We also investigate high-level database change operations that allow users to be concerned only with the new information they want to insert into the database and to leave the burden of resolving inconsistencies between the new and the old data to the database system. The principle of minimal change states that the result of adding the new information to a database should be the set of models of the new information that is closest to some possible models in the current database. Constructing similarity measure between spatial scenes allows us to study the change operations for constraint databases in terms of model-theoretic minimal change. We define several change operators based on the proposed similarity measure, and study the characteristics of these operators. Algorithms that implement the similarity queries and change operators are introduced. Analysis of the computational complexities of the algorithms shows that an efficient PTIME evaluation of similarity queries and several types of change operations in linear constraint databases is possible. We also describe and analyze experimental investigations on how well the similarity measures agree with human intuition. Finally, we describe the implementation of similarity queries within the MLPQ/GIS system.

Subject Area

Computer science

Recommended Citation

Deng, Ying, "Similarity queries in linear constraint databases" (2000). ETD collection for University of Nebraska-Lincoln. AAI9967362.
https://digitalcommons.unl.edu/dissertations/AAI9967362

Share

COinS