Abstract:

Attention Metadata (AM) have gained a great deal of interest in various communities such as Technology Enhanced Learning (TEL), Knowledge Work Support, Personal Information Management, Information Retrieval, Marketing and E-Commerce systems, and so on. The aim is to capture user interaction with many means ranging from server applications such as Learning repository, Learning Management systems (LMS), Content Management Systems (CMS) to applications in the user desktop such as e-mail, chat, instant messaging, and so on. We will focus our attention on the context of server applications in a learning environment. AM are collected with the aim to recommend new Learning Resources (LR) to users based on their current activity or goal. For a review of recommender systems in TEL the reader could refer to [1]. We propose the recommendation of query terms that can be used by the user to expand the search space on the basis of similarity at the query level or at the user level. The aim is to use the platform to collect the essential attention information that can be used to improve the user experience on the platform by recommending queries, that can be used to retrieve LR, and recommending users in order to facilitate the construction of a community of users with similar interests and a network of related LR.

The main objective of Learning Resources repositories (LRr) is to facilitate the sharing and reuse of resources. The usual starting point for reusing resources is a good search strategy, i.e. a good search condition or a good initial LR highly representative and highly referenced in the context of the query. The formation of a community of users with the same interest also greatly improves the process. The goal of collecting AM should be assisting the user in this process by recommending search conditions, LR, and users with a similar current interest.

The AM can be used to dynamically update users’ profiles in order to reflect the current activity and interest derived from the issued queries on the platform, the responses they received in the form of a list of LR, the use they made with the list of results in terms of which LR they selected, bookmarked, downloaded, repurposed. The dynamically updated profile can also be used to dynamically recommend new users with a current similar interest.

The query issued on the platform can be used to summarize the user’s activity by topic maps, e.g. topics covered by downloaded documents. The platform could give the possibility to use the topic map to self-evaluate the work and as a query expansion strategy.

The amount of AM collected should be as minimal as possible since too many or useless data produce noise and make the data mining process more difficult to perform. The AM should respond to who did what and when. Since the focus of the platform is to allow sharing and reusing LR, the first step for doing this is to assist the user in performing good searches. This could be done by guiding her in broadening the search space to facilitate innovation and by guiding the exploration of the result space by providing an efficient ranking mechanism to facilitate the synthesis. The AM should thus be built around the queries issued by the user to the systems. We will describe possible use cases of an LRr that can guide in developing the AM schema, metrics, services, and algorithms that could be provided by the platform. A brief review of the literature on the major AM schema will guide the discussion.

Keywords:

Attention Metadata, Learning Resource repositories, Ontological schema.