Research output: Book/ReportOther report (internal)


    Personalization of information is a key approach to overcome the plethora of information in the networked knowledge society. Personalization techniques rely on the combination of available datasets and information retrieval technologies such as data mining, collaborative filtering, or natural language processing. This combination can make thus far invisible pat-terns in the data visible and they can be used to create personalized recommendations for individual users. It is expected that educational datasets help determine the context of learners (individual needs, preferences, and learning goals) and thus personalize learning content and suggest suitable learning activities or peers to the learners. Personalized learning has the potential to reduce delivery costs, create more effective learning environments and experiences, accelerate time to competence development, and increases collaboration between learners. Recommender systems are a promising technology to realize personalized lifelong learning and are therefore increasingly applied in Technology Enhanced Learning (TEL). A lot of re-search has been conducted on recommender systems in TEL but it lacks mutually comparable ways to evaluate the performance of the different recommendation approaches and their effect on personalized (lifelong) learning. In the world of commercial recommender systems, it is common practice to use publicly available datasets with specific characteristics (e.g. MovieLens, Book-Crossing, or the EachMovie dataset) to evaluate recommender system al-gorithms. These datasets are used as benchmarks to develop new recommendation algorithms and compare their effects to other algorithms in a comprehensible way. Besides the evaluation of personalization technologies, datasets are increasingly used to create data mashups, that combine a number of data sources. Data mashups acquire their value from the data itself, and can create new insights and a new meaning through the combination of single data sources. A famous example for a data mashup is the Google Swine Flu epidemic mashup that spots the dissemination of the flue by analyzing searches that people were making in different regions of the world and showed these on a Google map. In TEL there is not one dataset publicly available, neither for the evaluation of recommender systems nor for the creation of data mashups.
    Original languageEnglish
    Publication statusPublished - 15 Dec 2010


    • NELLLeducational datasets
    • information retrieval
    • data mining
    • data science
    • data products
    • open data
    • data mashups
    • Learning Networks


    Dive into the research topics of 'AlterEgo'. Together they form a unique fingerprint.

    Cite this