Text attributes and processing techniques in geographical information systems
Mark J. Carlotto · International Journal of Geographical Information Systems · 1995
Text is an important data source in many GIS applications. In some applications, text data are in a format (e.g. tables) that can be directly converted into database records or objects. Other kinds of data such as historical records, field reports, and intelligence messages are unformatted and thus more difficult to incorporate into a GIS. Instead of simply attaching text to geographical data, an alternative paradigm is introduced where non-spatial data (free text as well as relational data) are converted into textual attributes. A textual attribute can be a free-text description of a geographically-referenced feature or event, relational data converted into a set of domain-specific terms, or a combination of domain-specific terms and free-text. An advantage of using textual attributes and text processing techniques to describe and retrieve geographical information is their extendibility. As the database changes over time and new attributes are required, attribute-value pairs can be converted into text attributes and added to the system without otherwise altering the database. A vector-based representation is used to efficiently encode text attributes. Such a scheme is well-suited for incorporating text data into an object-oriented GIS. Text vectors are stored in slots local to the objects they describe. As new data are added to the system, existing vectors are not affected. Thus the database is easy to maintain. Methods can be easily implemented within the object-oriented framework to measure the similarity of object vectors to the vector of another object or a query. The similarity or score can be combined with other similarity measures (e.g. spatial/temporal proximity) to allow geographical information to be retrieved by location, time, and content in a uniform fashion. Two case studies are presented to illustrate the usefulness of textural attributes and text processing techniques in describing, accessing, and visualizing geographical information.