System Architecture - Technologies

ShoppingMate General Architecture

The ShoppingMate environment is an integrated system that provides a number of advanced services to customers. It consists of a number of components that interact with each other. Each component is responsible for a specific part of the information workflow. The key components of ShoppingMate are the user interface, the shop interface, the semantic representation engine, the recommendation engine and the advertiser, as shown in the above figure.

The user can utilize an intelligent and efficient interface that provides all the necessary facilities in order to collect and present the required information in the framework of an interaction with the ShoppingMate system. This interface is lightweight because mobile terminals have not increased computational capabilities. Menu based selection and free text are the tools for the insertion of user information. This information is represented by ontological terms in the Intermediate Server that is responsible for the manipulation of the semantic representation. It should be noted that both the user profile and user requests are represented by ontological terms providing a lot of advantages in the interaction process. For example, ontological representation can be used for personalized services provision. Hence, the user starting his interaction with the system defines personal details as well as his preferences that are described by ontological terms in the application server. Afterwards, he can place his queries to the system asking for products that best match his preferences. The Intermediate Server is responsible to translate the user query by using the defined ontology and accordingly to retrieve the query results from the recommendation engine. This way, efficient and personalized results are retrieved increasing the user satisfaction level.

Another important component of the ShoppingMate system is the shop agent. The shop agent is responsible to reflect the status of shops and their items. A flexible interface is used for allowing the shop owners to insert and manipulate information related to shop items. In parallel, automated capture procedures will be explored and to the shop owner. This information is also represented by ontological terms. Specific instances are created in the shop ontology storing basic information for items such as colour, size, price, etc. The most important is that a process for the automatic ontology creation based on database descriptions for items will be available in the ShoppingMate framework. Efficient techniques will be used for these purposes.

The semantic representation of critical information will be the base for the recommendation engine in order to provide personalized services in an efficient way. The recommendation engine utilizes machine learning technologies and it is responsible for producing scored matches for users, shops and items. Recommendations are based on an advanced set of criteria, such as preferences, interests, behavior and on learning techniques of user behavior patterns. Patterns derived from monitoring the user behavior are added on-the-fly during system operation. The recommender service is accessed via a flexible query language (SMART Recommender Query Language, SQL) that allows requesting context specific recommendations on a detailed level.

The advertiser is a mechanism for sending advertisements for the benefit of the user and not overloading her/him with annoying content. Based on a user request, this component will send targeted advertisements about items that the user may be interested in. These advertisements will be available both on demand (following a user’s selection) and also as the consumer comes in short range with the corresponding shop. Provided the user’s permission, the system will anonymously monitor, identify, and analyze user’s actions while shopping, thus enabling routine learning and optimizing its functionality. The usage of the advertiser makes the ShoppingMate system a productive and valuable new marketing tool that provides targeted advertising facilities.

Users’ community service allows users to rate products for their quality, price, etc. and shops for their reliability, friendliness, etc. This is very important because the ratings can be very useful when a user decides to buy a specific product. Ratings can be taken into consideration when the system decides which products will be included in the returned list of products. Moreover, users can review products or shops. The community service will be fully integrated in the user-client and will allow users to send messages to each other. The users’ community service is strongly coupled with the recommender and the advertiser. The ShoppingMate application uses Web Service or REST based interfaces provided by the ShoppingMate community service to provide community related features in the application.

The navigator is a component designed to assist the user as she/he moves from shop to shop. Taking as input the outcome of the recommender or advertiser components, its mission is to propose a specific route for the movement of the user from shop to shop. It can use a map-based approach and provide route optimization facilities. This way the user will be guided more efficiently when there is the need for movement to another shop.