A Context-based Mobile Knowledge Support System for Supporting Foreigners: Demand-Supply Knowledge Representation and Matching Algorithm
【摘要】：In the past few years, the increasing popularity of mobile devices and the rapid development of the wireless networking technologies are enabling some new classes of applications in various fields such as education, health, tourism, and so on. These applications were marked by a special feature of movement. The progress made in semantic web and ubiquitous computing has led to the development of mobile applications which can provide adapted services according to the user's requirements such as, context-aware system for overcoming information overload problem. Knowledge support system is one of the applications which are characterized by precision, rapid provision, right time, and right place, as well as adding the feature of mobility and personalization to be context-based mobile knowledge support system for supporting foreigners in foreign countries. Foreigners in foreign countries, who are there for studying, working or living, may face various challenges that can be divided into three main categories, namely knowledge needs, language skills, and guider, which are related to the differences in cultures, languages and the geographic information respectively. For example, in recent years, there are a large number of foreign students coming to China to learn Chinese language, work, or live. However due to the characteristics of the local culture, Chinese language complexity, and the geographical complexity of many areas, it is usually difficult for them to do any activity without human guider.
In this research, the current available technologies allows for their exploitation creation of an approach called Context-based Mobile Knowledge Support System (CMKSS) for supporting foreigners in foreign countries, which will be considered as a contribution for solving the problems faced by foreigners. CMKSS approach provides three types of supports, namely knowledge support, language support and guide support, when the foreigner wants to fulfill any activity in a foreign country. In order to fulfill an activity, firstly he/she needs some knowledge about the specific task which allows him/her to make some decisions, secondly he/she needs to learn some language skills to communicate with the locals, and finally he/she needs some geo-information to guide him/her to achieve this task. From the theoretical perspective, the real value of CMKSS is the efficient modeling of the changeable context information and the fast matching process. This research will discuss the following three questions:
Q1:How to represent the user's requirements in a good form and make it easy to get what the user exactly wants? Due to problems exist in keyword-based searching, which leads to the misunderstanding of the user's needs and the problem of dealing with the changeable contextual information of the user's situation and the surrounding environment. The difficulties are (1) How to recognize the user's needs which inputted as free form query. (2) How to obtain the changeable contextual information and represent these contexts for constructing the user's demand profile, which reflects a user's needs.
Q2:How to represent the support item in a proper form to make it easy to be retrieved and fit the user's requirements? Traditional methods of indexing the support items are insufficient for providing semantic matching. The difficulty is how to establish the semantic relations to accurately defining the contents of the support item in order to enhance the semantic matching and the context based matching.
Q3:How to efficiently match user's requirements with the suitable support items to fit his/her needs? Due to both low efficiency and effectiveness of the classical exact matching methods and the limited capabilities of the mobile device, which may lead to the mismatching or no-matching problems, the difficulty is how to achieve the approximate match between the changeable contextual information of the user's requirements and the available support items to provide top-n suitable support items, which fit the user's needs and surrounding environment. In this research, the processing of semantic query and ontology-based context model are proposed to represent the user's requirements, three types of context categories, namely activity contexts, device contexts and user contexts are used in this model to accurately define the user's needs composed into the demand profile that is represented by an XML file. Semantic annotation mechanism is proposed to annotate the support items stored in the document's repository. WordNet-ontology is used in word sense disambiguation for extending the concepts in the support item description. Domain-ontology is used for annotating the support items with the domain ontology's concepts and then supply profile is represented by RDF graphs. To match the top-n suitable support items, the proposed matching algorithm consists of three parts. The first part is activity-based semantic ranking to rank the support items related to the activity contexts. The second part is device-based filtering to filter the support items which are run able in the device used. The third part is user-based top-n matching to select the top-n support items, which fits the user's preferences.
The contributions of this research are as follows. The first contribution is an ontology-based knowledge representations for the demand knowledge of acquiring the user's requirements and the supply knowledge of annotating the support items are proposed. The second contribution is an efficient and effective matching algorithm, including semantic matching and context-based matching, is proposed to approximately match between the supply profiles and the demand profile. The third contribution is A mobile knowledge support system for foreigners is developed for supporting them in foreign countries with knowledge, language, and guide supports. The knowledge representation and the matching algorithm can be considered as the answers to the scientific questions mentioned above. The final results show that CMKSS can efficiently provide the top-n suitable support items for a user based on the user's needs and surroundings.