On the Problem of Logical Recognition in the Dynamic Internet Environment

The problem of formalized logical recognition in conditions of permanent modification of information content on the Internet is considered. An approach based on the idea of modifiable reasoning is proposed. As a mathematical technique for this approach, we propose to use the method of logical actors. This method is compared with the nonmonotonic reasoning. Received October 25, 2000 INTRODUCTION The complexity of the retrieval and recognition of information entities on the Internet is caused by the following reasons. 1. The information on the Internet has a complex structure permanently being renovated. 2. The space of informational search is extremely large. Since the features of sought for entities have a complex structure, it is expedient to use logical models and languages when solving the problems of information retrieval and recognition in the Internet environment. However, the use of logical reasoning is substantially complicated by the necessity of ensuring the soundness and completeness of logical inferences under conditions of permanently updating information on the Internet. The data may undergo changes in the course of the execution of a logic program, which may result in recognition errors. In this paper, we discuss the problems of developing a mathematical tool for the modifiable reasoning that would ensure the validity of logic programs (intelligent agents) that function in the dynamic Internet environment, where the informational content is permanently changing and augmented. We compare the nonmonotonic reasoning and the method of logical actors as possible bases for the tool mentioned above. 1. THE MODEL OF AN INTELLIGENT AGENT Consider a simplified model of an intelligent agent that performs the information retrieval and recognition on the Internet (see figure). Let us assume that our agent is a certain logic program that processes data extracted from the Internet and outputs reports on the gathered data. In the context of such a model, the following questions are of vital importance: 1. Does the intelligent agent work with certain Internet resources specified in advance or do the resources depend on the incoming data and are they selected in the course of program execution? 2. How long should this agent operate? Certain agents are used only occasionally to carry out one-time search operations on the Web (by the way, this task is quite adequately executed by conventional search engines like Rambler, Alta Vista, etc.), while others continually operate for months or even years. Obviously, when an agent works with certain resources specified in advance or when it is used for one-time search operations, the new report based on the SOFTWARE AND HARDWARE IN SYSTEMS OF PATTERN RECOGNITION AND IMAGE ANALYSIS