Innovative Mining Technologies and Information Recommendation for Electronic Services and Commerce

With the emergence of Internet access around the world, consumer choices have multiplied exponentially. Even if this allows for an abundance of options and a wider variety of choice, it has become increasingly difficult to match consumer preferences with the most suitable products, especially considering the diversity of needs among them. Recommender Systems attempt to regulate this by analyzing consumer preferences and predicting the user's preference for a new item. This is primarily achieved by analyzing the user's existing known preference-ratings for certain items, and then predicting the user's preferences for new items, based on the aforementioned analysis. The ultimate goal of this proposal is to design and develop a Recommender system that incorporates the latest technologies in the field, as well as a wealth of innovations, making it capable of addressing many known issues in the field, as well as making it applicable to a wide range of different cases. The innovations of the proposed system include the use of synthetic coordinates, community detection methods, Cloud Computing technologies and a new approach to creating a hybrid system.


The new demand of the Information Age is the optimal management of the huge amount of available information, with the basic embodiment of this management being the identification of information related (conceptually and semantically) to the needs of each individual. The object of the project is Information Recommendation Systems, a basic and valuable tool for achieving this goal. These systems aim to solve the problem of recommending content to users, which can be described as follows: Given a set of users and a set of objects (content), an estimate (prediction) of the rating of any object from any user needs to occur. In the sub-problem of personalized recommendation, given a set of known ratings of said user towards some objects, the goal is to provide recommendations that more accurately reflect the specific user's preferences, as they emerge from existing ratings.

The objectives of the project are divided into three main categories, covering the entire range of activities of the Business Partnership with Research Organizations:

  1.  Research, design, development, and evaluation of a prototype hybrid algorithm for providing personalized recommendations. Characteristics-targets of this research result are:

    a. Providing highly accurate recommendations in an efficient manner, surpassing existing algorithms in accuracy.

    b. Its applicability to a wide range of applications and operating environments.

    c. The high degree of scalability, that is, the ability of the algorithm to provide high-level results, in an efficient way, for an extremely large number of users and objects.

    d. Addressing known (stated in bibliography) problems of existing recommendation systems, such as the "Cold start", "Filter Bubble" and "Popularity-bias" problems.

e. Analyzing system usage data and detecting patterns and key factors in user preferences.

  1. The design, development, and implementation of a (based on the aforementioned algorithm) prototype real tool for information recommendation, easily integrable in a wide range of real use scenarios, such as online stores, content delivery systems, social networking, tourist services, etc. Key features of the tool will be:

a. The recommendation of information, regardless of parameters such as type of objects, evaluation format and operating platform. The main goal is the easy adoption of the tool in the largest possible number of usage scenarios.

b. Comprehensive analysis and presentation function of usage data to administrators, such as profiles of user choices, profiles of popular item attributes, as well as detailed prediction of new items' appeal to the user base, aimed at more effective strategy planning.

  1. Diffusion of the work produced, as well as the usefulness of information recommendation systems in the private sector with the appropriate actions to inform all interested parties, as well as the pilot but also actual implementation of the system in collaboration with interested companies.


Send us your cv

Maximum allowed file size is 4MB

Fields marked with an asterisk (*) are required