Manchester Metropolitan University's Research Repository

    Arabic goal-oriented conversational agent based on pattern matching and knowledge trees

    Noori, Z, Bandarl, Z and Crockett, K (2014) Arabic goal-oriented conversational agent based on pattern matching and knowledge trees. In: World Congress on Engineering 2014.


    Download (1MB) | Preview


    Conversational Agents (CA's) are computer agents used in applications to converse with humans using natural language dialogues. They are widely used in different fields like industry, education, marketing, health, and other services. Goal Oriented Conversational Agents (GO-CAs) are agents having a deep strategic purpose which enables them to direct conversations to achieve a certain goal using a specific domain. Typically (CA's) are programmed to have a set of rules that guide the conversation with the user. One technique used to script CA's is through pattern matching algorithms. Such algorithms are used to match the user's dialogue and instigate the conversation through writing a series of scripts that contains the rules and patterns relevant to the domain. Throughout the conversation, values can be extracted from the user's dialogue which allows the CA to respond with the correct answer. CA's have been mainly developed for the English language and very limited work has been carried out in Arabic. This is mainly due to the complexity of the language and the lack of resources supporting the Arabic language. This paper proposes a new CA architecture based on a pattern matching algorithm for the development of a goal orientated Arabic Conversational Agents (ACA). The ACA incorporates a new scripting language and knowledge engineering is used to construct the domain. A prototype ACA was developed and the Iraqi passport system was used as a domain to evaluate the new ACA. The ACA was tested and evaluated by experts within the Iraq Consulate with encouraging results and received positive feedback.

    Impact and Reach


    Activity Overview
    6 month trend
    6 month trend

    Additional statistics for this dataset are available via IRStats2.


    Repository staff only

    Edit record Edit record