Qiming Chen

A structured approach for cooperative query answering

This paper proposes the use of a type abstraction hierarchy as a framework for deriving cooperative query answers. The type abstraction hierarchy integrates the abstraction view with the subsumption (is-a) and composition (part-of) views of a type hierarchy. Such a framework provides multilevel object representation, which is an important aspect of cooperative query answering. The concept of pattern that specifies one or more conditions on an object is also proposed. Patterns have smaller granularity than types, and thus provide more specific semantic information. Cooperative query answering consists of query relaxation, generalization, specialization, and association on patterns. Query relaxation can be explicitly specified by the user or implicitly performed by the system. The implicit and explicit relaxations can also be combined and performed interactively by both the system and the user. CSQL, an extension of SQL for cooperative query answering, is also proposed. Preliminary experimental results reveal that the proposed type abstraction hierarchy provides an organized structure representing concepts at different knowledge levels in various domains, and provides a systematic and efficient method for cooperative query answering. >

Neighborhood and associative query answering

Cooperative query answering extends the classical notion of query answering to provide neighborhood and associated information. Neighborhood query answering relaxes the query and its answer via abstract representations. To integrate the abstraction view with the subsumption (is-a) and composition (part-of) views of type hierarchy, the notion of type abstraction hierarchy is introduced. To evaluate and control query relaxation, a nearness measure mechanism is provided. Associative query answering provides information conceptually related to, but not explicitly asked by the query. As object association is context sensitive, a DB-Pattern-KB framework is developed that couples domain-specific knowledge and participating objects in localized problem domains via virtual database patterns. Associative query answering can then be accomplished through tracing the behavior dependencies among cooperating objects in those problem domains. Such a framework allows related databases and knowledge bases to be linked dynamically in various contexts yet be maintained relatively independent of each other. The proposed approach has been implemented in the cooperative database system tested, CoBase, at UCLA. Our experience reveals that the proposed techniques are effective for cooperative query answering.

Cooperative Query Answering via Type Abstraction Hierarchy

Cooperative query answering consists of analyzing the intent of the query and providing generalized, neighborhood or associated information relevant to the query. The key issues to accomplish cooperative query answering consist of supporting different knowledge representations at different abstract levels and providing inference between these levels. In this paper the Type Abstraction Hierarchy is proposed which is characterized by dealing with subtyping from subsumption, composition, and also abstraction views. Based on the type abstraction feature provided by this model, an inference technique for cooperative query answering is developed. Such an inference is performed by abstracting and refining the goal to generalize and specialize the query scope and to derive relevant answers with different generality, coverage, and approximation, or to link related subjects at certain levels by using different representations of knowledge given at different levels. A prototype system has been implemented at UCLA that demonstrates the use of this approach for such decision making and problem solving applications as conceptual query processing, neighborhood inference, and subject association.

Using type inference and induced rules to provide intensional answers

A new approach is presented that uses knowledge induction and type inference to provide intensional answers. Machine learning techniques are used to analyze database contents and to induce a set of if-then rules. Type inference which is based on forward inference and backward inference is developed that uses database type hierarchies to derive the intensional answers for a query. It is shown that more precise intensional answers can be derived by properly merging the type inference results from multiple type hierarchies. A prototype intensional query-processing system which uses the proposed approach has been implemented. Using a ship database as a testbed, the effectiveness of the use of type interference and induced rules to derive specific intensional answers is demonstrated.<<ETX>>

A pattern based approach of integrating data and knowledge to support cooperative query answering

A framework for integrating class-oriented data grouping and subject-oriented knowledge grouping to support cooperative query answering is proposed. This framework is characterized by a three layer data/knowledge organization: object layer, object-subject layer, and subject layer, where a database at the object layer and a knowledge base at the subject layer are coupled by virtual database patterns specified at the object-subject layer. Based on such an architecture, communicating databases and knowledge bases are organized and maintained independently but linked dynamically under specific subjects. Cooperative query answering can then be accomplished through tracing the behavior dependencies amongst cooperating objects under those subjects. To implement this approach, mechanisms of dynamic-classifications, deductions, focus transitions and goal rewrites are introduced. An experimental cooperative database system, CoBase, was developed to demonstrate the effectiveness of the approach.<<ETX>>

Query answering via cooperative data inference

This paper proposes the use of accessible information (data/knowledge) to infer inaccessible data in a distributed database system. Inference rules are extracted from databases by means of knowledge discovery techniques. These rules can derive inaccessible data due to a site failure or network partition in a distributed system. Such query answering requires combining incomplete and partial information from multiple sources. The derived answer may be exact or approximate. Our inference process involves two phases to reason with reconstructed information. One phase involves using local rules to infer inaccessible data. A second phase involves merging information from different sites. We shall call such reasoning processes cooperative data inference. Since the derived answer may be incomplete, new algebraic tools are developed for supporting operations on incomplete information. A weak criterion called toleration is introduced for evaluating the inferred results. The conditions that assure the correctness of combining partial results, known as sound inference paths, are developed. A solution is presented for terminating an iterative reasoning process on derived data from multiple knowledge sources. The proposed approach has been implemented on a cooperative distributed database testbed, CoBase, at UCLA. The experimental results validate the feasibility of this proposed concept and can significantly improve the availability of distributed knowledge base/database systems.

HILOG: A HIGH-ORDER LOGIC PROGRAMMING LANGUAGE FOR NON-1NF DEDUCTIVE DATABASES

A formal framework of a strongly typed logic programming language with high-order terms (HILOG) is developed which extends Logic Programming (LP) from First-Order Logic (FOL) and which generalizes deductive databases to handle non-normalized relations. To remedy the limitations of current approaches by treating complex objects as function terms of the FOL, we reformulate the key Logic Programming (LP) notions through the introduction of appropriate mathematical concepts such as partial containment, packing and unpacking. In this paper, we have developed the extended notion of the satisfaction, the existence of a minimal model closure, the uniqueness of a standard (packed) minimal model for a HILOG program, the model p-intersection theorem, and the extended least fixpoint characteristics of the HILOG minimal model. Therefore, HILOG provides a canonical framework for high-order LP in which semantics covers the key points of LP. HILOG Language has the capabilities of representing structured knowledge and type hierarchies which are important for integrating the LP notions to other programming systems for handling complex objects and for developing non-1NF deductive databases. An example is given to demonstrate the use of HILOG in representing structured knowledge and in deductive retrieval of complex objects.

Fault tolerant distributed database system via data inference

A knowledge-gased approach for query processing during network partitioning is proposed. The approach uses available domain and summary knowledge to infer inaccessible data to answer a given query. A rule induction technique is used to extract correlated knowledge between attributes from the database contents. This knowledge is represented as rules for data inference. On the basis of a set of queries, simulation is used to evaluate the effectiveness of the proposed data inference technique for improving data availability under network partitioning. Object allocation has a significant impact on data availability. Allocating objects that increase remote redundancy and reduce local redundancy increases data Availability during network partitioning. A prototype distributed database system that uses the proposed inference technique with correlated knowledge from a ship database has been implemented. Experience indicates that the proposed inference technique can significantly improve the availability of a distributed database during network partitioning.<<ETX>>

A pattern-based approach for deriving approximate and intensional answers

A pattern-based approach is proposed to derive the approximate and intensional query answers when the exact answer is unavailable or too time-consuming to generate. The approximate and intensional query answers may be refined if more time is available. Since the pattern-based query processing performs mainly main-memory based manipulations without database access until the last step of generating the final results, it should provide faster response to queries than the conventional query processing in real-time applications.<<ETX>>

Pattern-Based Data and Knowledge Integration for Intelligent Query Answering

A framework for integrating c1ass-oriented data groupings and subject-oriented knowledge groupings to support intelligent query answering is proposed. This framework is characterized by a three-layer data/knowledge organization: object layer, object--subject layer, and subject layer, where a database at the object layer and a knowledge base at the subject layer are coupled with virtual database patterns specified at the object--subject layer. In such an architecture, communicating databases and knowledge bases are organized and maintained independent of each other, but linked dynamically under specific application domain and context. The proposed approach avoids object duplication, migration, and respecification, thus providing a flexible and scalable technique for integrating knowledge base and database. An experimental cooperative database system, CoBase, has been developed at UCLA to demonstrate this approach in supporting intelligent query answering and in integrating large scale databases and knowledge bases.