Semantic Engine

Framework for semantic recommendations

Kompetenzzentrum:  Information Retrieval and Machine Learning
KontaktProf. Dr.-Ing. Sahin AlbayrakDr.-Ing. Andreas Lommatzsch
Website: --

 

Introduction

With the growing availability of semantic data, the processing of this data comes more and more in the focus of interest. Especially the efficient processing of large semantic datasets promises an improvement of different applications in many domains. The Semantic Engine project introduces an architecture that supports the aggregation of different types of semantic data and provides components for deriving recommendations and predicting relevant relationships between dataset entities. The developed architecture supports different types of data sources (e.g. databases, semantic networks) and enables the efficient processing of large semantic datasets with several different semantic relationship types. For demonstration the capabilities of the developed framework, we develop application for the entertainment domain.

The deployment of semantic approaches for information filtering provides the basis of handling multi-lingual data sources and for providing highly relevant information. Semantically represented knowledge is extracted from unstructured data and interlinked with existing semantic data collections (e.g. Freebase). We analyze the created comprehensive knowledge base using many different advances machine-learning approaches, such as clustering and graph kernels. With the overwhelming availability data, the filtering and management of large data collections becomes one of the most critical issues in the daily life. Relevant data is often distributed among different sources and data formats preventing the easy access and the extraction of relevant information.

Motivation

Current information systems usually are designed to compute results only for one predefined scenario. These systems are often based on a fixed processing pipeline, supporting only one data source and one recommender algorithm. By contrast, the approach described the Semantic Engine defines a universal framework, supporting different data stores (e.g. databases, RDF triple stores) and several state-of-the-art recommender algorithms. The framework is open to new graph models and supports the customization to the specific requirements for the respective scenario. A specialized component allows the context-aware aggregation of results computed based on different semantic models.