Peretz Shoval

Information Filtering: Overview of Issues, Research and Systems

An abundant amount of information is created and delivered over electronic media. Users risk becoming overwhelmed by the flow of information, and they lack adequate tools to help them manage the situation. Information filtering (IF) is one of the methods that is rapidly evolving to manage large information flows. The aim of IF is to expose users to only information that is relevant to them. Many IF systems have been developed in recent years for various application domains. Some examples of filtering applications are: filters for search results on the internet that are employed in the Internet software, personal e-mail filters based on personal profiles, listservers or newsgroups filters for groups or individuals, browser filters that block non-valuable information, filters designed to give children access them only to suitable pages, filters for e-commerce applications that address products and promotions to potential customers only, and many more. The different systems use various methods, concepts, and techniques from diverse research areas like: Information Retrieval, Artificial Intelligence, or Behavioral Science. Various systems cover different scope, have divergent functionality, and various platforms. There are many systems of widely varying philosophies, but all share the goal of automatically directing the most valuable information to users in accordance with their User Model, and of helping them use their limited reading time most optimally. This paper clarifies the difference between IF systems and related systems, such as information retrieval (IR) systems, or Extraction systems. The paper defines a framework to classify IF systems according to several parameters, and illustrates the approach with commercial and academic systems. The paper describes the underlying concepts of IF systems and the techniques that are used to implement them. It discusses methods and measurements that are used for evaluation of IF systems and limitations of the current systems. In the conclusion we present research issues in the Information Filtering research arena, such as user modeling, evaluation standardization and integration with digital libraries and Web repositories.

Principles, procedures and rules in an expert system for information retrieval

Abstract An expert system was developed in the area of information retrieval, with the objective of performing the job of an information specialist, who assists users in selecting the right vocabulary terms for a database search. The system is composed of two components: One is the knowledge base, represented as a semantic network, in which the nodes are words, concepts, phrases, comprising a vocabulary of the application area and the links express semantic relationships between those nodes. The second component is the rules, or procedures, which operate upon the knowledge-base, analogous to the decision rules or work patterns of the information specialist. Two major stages comprise the consulting process of the system: During the “search” stage relevant knowledge in the semantic network is activated, and search and evaluation rules are applied in order to find appropriate vocabulary terms to represent the users problem. During the “suggest” stage those terms are further evaluated, dynamically rank-ordered according to relevancy, and suggested to the user. Explanations to the findings can be provided by the system and backtracking is possible in order to find alternatives in case some suggested term is rejected by the user. This article presents the principle, procedures and rules which are utilized in the expert system.

Stereotypes in information filtering systems

Abstract A stereotype represents a collection of attributes common to a cross-section of people. The use of stereotypes is efficient in information systems that need to model their users in order to improve the interaction between users and the system. Expert systems and various tutors have used stereotypes as a helpful means of modeling users. Stereotypes enable systems to make plausible inferences about users on the basis of their stereotypic association. This paper discusses the use of stereotypes as a means of improving the effectiveness of information filtering systems that are based on user profiles. We describe various approaches of using stereotypes in existing systems, and propose a unique way of using stereotypes in a new model for information filtering.

Database schema design: an experimental comparison between normalization and information analysis

This article compares two different methods for designing a data base schema-normalization and information analysis (IA). A set of design tasks was assigned to two groups of analysts who were trained to use the two methods in conjunction with the structured analysis method of system analysis. The results of the experiment revealed that the quality of the data base schemata designed using normalization was better than that designed using IA, that normalization required less time than IA to perform, and that the analysts preferred normalization.

ADISSA: architectural design of information systems based on structures analysis

Abstract ADISSA is a methodology for Architectural Design of Information Systems and Software, based on Structured Analysis. The architectural design includes: 1. (a) design of a tree-structured menu system, which interfaces between the users and the system—viewed as the “external architecture” of the system; and 2. (b) design of the transactions of the system, composed of various functions which are activated in response to user needs and various events in the universe of discourse—viewed as the “internal architecture” of the system. The menu tree and the transactions are tied up with the other major components of the system architecture: the inputs and outputs, and the database. ADISSA methodology is fully compatible and consistent with Structured Analysis, and therefore the stages of system analysis and design are integrated into one complete process.

Database reverse engineering: from the relational to the binary relationship model

Abstract This paper describes an algorithmic method for transforming a relational database schema to a binary-relationship one. The source schema may consist of relations that are at any level of normalization, and the designer may add semantic information on the source schema, such as the definition of candidate keys, foreign keys, functional dependencies of various types, multi-valued dependencies, many-to-many constraints, inclusion dependencies, and others. Based on this information, the multi-stage transformation algorithm applies mapping rules to generate object-types, binary-relationships and constraints in the target conceptual schema. The method is implemented as a PC-based tool, utilizing Ingres, SQL and C, and is part of a comprehensive database design tool for both forward and reverse engineering.

Binary-relationship integration methodology

Abstract BRIM is a view integration methodology which utilizes the binary-relationship data model and encompasses all stages of conceptual schema integration. In the pre-integration stage it assists the designer to prepare a binary-strategy integration plan. Then, in the stage of conflict resolution, it identifies and resolves all possible types of conflict between schemas, including naming conflicts (i.e. homonyms and synonyms) and structural conflicts (i.e. conflicts regarding differences between types, dependencies, keys, constraints, and hierarchies of objects). Once conflicts are resolved and the source schemas are in conformity, they are merged by superimposition, and inter-schema relationships are added to the integrated schema. The process is repeated for every two source schemas or intermediate results, until the target global schema is obtained. BRIM is supported by an interactive software tool which is part of a PC-based system for automated database design. The tool leads the designer through all stages of the integration process, applying conflict resolution rules and schema merging algorithms, providing an integrated conceptual schema which can then be automatically tranformed into a normalized database schema.

FOOM - Functional and Object-Oriented Analysis and Design of Information Systems: An Integrated Methodology

We propose FOOM (Functional and Object-Oriented Methodology), an integrated methodology for information systems analysis and design, which combines two essential software-engineering paradigms: the functional/data approach (or process-oriented) and the object-oriented (OO) approach. System analysis phase, where user requirements are set and defined, includes functional analysis and data modeling activities. This phase produces a hierarchy of data flow diagrams (DFD) and an initial OO schema, which can be created directly or from an entity-relationship diagram (ERD). The design phase is performed according to the OO approach, producing a complete OO schema and a behavior schema. The seamless transition from analysis to design is enabled thanks to ADISSA methodology, which facilitates the design of the menus, forms and reports classes, and the system behavior schema, from the DFDs and the application transactions. The paper introduces the motivation for the combined approach, outlines the methodology, and presents an example that demonstrates it.

Experimentation with an information filtering system that combines cognitive and sociological filtering integrated with user stereotypes

Abstract A dual-method model and system for filtering and ranking relevance of information is presented. One method is cognitive filtering, while the other is sociological filtering, which is integrated with user stereotypes. A prototype system was developed to test the applicability of the model for filtering e-mail messages, and experiments were run to determine the effects of combining the two methods in various filtering strategies. Results reveal that although cognitive filtering alone is usually more effective than sociological filtering alone, the combination of both methods yield better results than using each method individually. Ordinarily, the best filtering strategies are achieved when the two methods are used in parallel, or when cognitive filtering is the primary method, followed by sociological filtering. We conclude that the optimal filtering strategy of combining cognitive and sociological filtering is stereotype dependent; i.e., for each user stereotype, there may be a specific combination of the cognitive and sociological filtering that yields best results.

ADDS: A system for automatic database schema design based on the binary-relationship model

Abstract This paper presents the system ADDS that has been developed to assist the database designer designing a database schema. A distinction is made between the stage of information structure analysis in which the information structure of the system is defined according to its user information needs, and the stage of database schema design in which the record types of the database and the relationships between them are defined. In the first stage a conceptual schema is obtained, represented as an information structure diagram (ISD), and in the later stage the ISD is used to derive the database schema in the form of a data structure diagram (DSD). ADDS automatically creates the database schema out of a conceptual schema which is expressed as an ISD of the binary-relationship data mode. The resulting schema consists of normalized record types, according to the relation model, along with hierarchical/set relationships between ‘owner’ and ‘member’ record types, as in the CODASYL/Network model. ADDS applies algorithms to convert the conceptual schema into the database schema. It is implemented on a micro-computer under MS-DOS using dBASE III.