University of Nairobi

A world-class university committed to scholarly excellence

A-Z List of UON Departmets
a b c d e f g h i j k l m n o p q r s t u v w x y z 

Home > Githiari Lawrence Muchemi > My Publications

My Publications

Publications (3 Records)

  • Towards Full Comprehension of Swahili Natural Language Statements for Database Querying - 2008

    ABSRACT.

    Natural language access to databases is a research area shrouded by many unresolved issues. This paper presents a methodology of comprehending Swahili NL statements with an aim of forming corresponding SQL statements. It presents a Swahili grammar based information extraction approach which is thought of being generic enough to cover many Bantu languages. The proposed methodology uses overlapping layers which integrate lexical semantics and syntactic knowledge. The framework under which the proposed model works is also presented.  Evaluation was done through simulation using field data on corresponding flowcharts. The results show a methodology that is promising.

       ICCR

  • Enhancing Citizen-Government Communication Through Natural Language Querying - 2007

    ABSRACT. The fast growth of computerized data processing and communications, and the desire by most African governments to introduce e-governance has brought great hopes for efficiency in public sector information dissemination. This has brought a host of challenges and key among them is in the area of data retrieval from relational databases. Most members of the general public in East Africa, and indeed the rest of the world, are not very conversant with database manipulations using form based interfaces and therefore may not take full advantage of computerized government information systems in the eventuality e-governance takes off. Swahili natural language querying (NLQ) is proposed as the most effective way of solving this problem. However there are some key unresolved issues hindering this solution.

     

    This paper reports on these key outstanding research issues together with theoretical and methodological approaches being applied to Swahili natural language querying (NLQ). The issues include lack of computerized phrase identification mechanisms, difficulties in computerized information extraction using traditional syntactic processing methods and non-existence of developed Swahili NL query translation algorithms to SQL. The paper proposes how to surmount some of these challenges and possibly achieve a practical way of data retrieval from relational databases for Swahili users. It proposes the use of meaning representations which can be implemented as a frame-based prototype.

     

    Keywords: natural language querying (NLQ), meaning representations, frames

       1st International Conference in Computer Science and Informatics (COSCIT 2007, UoN-ISBN 9966-7284-0-6, Nairobi, Kenya 150-154

  • SEMANTIC BASED NL FRONT-END FOR DB ACCESS: FRAMEWORK ADOPTATION FOR SWAHILI LANGUAGE - 2007

    ABSRACT: The forty years of Natural Language (NL) processing research have not reached the necessary threshold to move results from laboratories to practical realities of commercial world. This unsatisfactory state in NL processing has been a result of an erroneously chosen philosophy and methodology that considers human language understanding as a syntactic problem while significantly ignoring semantics and pragmatics aspects of NL. These have a key role in comprehension of NL texts within narrow subject domains such as natural language queries (NLQ) to databases.

    In this paper we consider some NLQ paradigm which integrates the following principles:

    A.     Semantically-oriented approach to NL analysis.

    B.    Efficient use of a problem-oriented knowledge within the NL analysis/understanding process

    C.   A bottom-up distributed self-organizing parsing which would be based on data-driven type of control.

    The NLQ paradigm is presented as a framework. Evaluation has been done by development and testing of a prototype resulting from the above framework. The prototype has been successfully tested for English and Russian NLQ. Through examples we show that this framework can successfully be adopted for Swahili NLQ as well.

     

    Keywords: natural language querying, pragmatics, semantics, parsing

       COSCIT 2007

[First Page] [Prev] Showing page 1 of 1 pages [Next] [Last Page]