Stephan Baumann

DOLCE ergo SUMO: On foundational and domain models in the SmartWeb Integrated Ontology (SWIntO)

Increased availability of mobile computing, such as personal digital assistants (PDAs), creates the potential for constant and intelligent access to up-to-date, integrated and detailed information from the Web, regardless of ones actual geographical position. Intelligent question-answering requires the representation of knowledge from various domains, such as the navigational and discourse context of the user, potential user questions, the information provided by Web services and so on, for example in the form of ontologies. Within the context of the SmartWeb project, we have developed a number of domain-specific ontologies that are relevant for mobile and intelligent user interfaces to open-domain question-answering and information services on the Web. To integrate the various domain-specific ontologies, we have developed a foundational ontology, the SmartSUMO ontology, on the basis of the DOLCE and SUMO ontologies. This allows us to combine all the developed ontologies into a single SmartWeb Integrated Ontology (SWIntO) having a common modeling basis with conceptual clarity and the provision of ontology design patterns for modeling consistency. In this paper, we present SWIntO, describe the design choices we made in its construction, illustrate the use of the ontology through a number of applications, and discuss some of the lessons learned from our experiences.

Using cultural metadata for artist recommendations

Our approach to generate recommendations for similar artists follows a recent tradition of authors tackling the problem not with content-based audio analysis. Following this novel procedure we rely on the acquisition, filtering and condensing of unstructured text-based information that can be found in the Web. The beauty of this approach lies in the possibility to access so-called cultural metadata that is the agglomeration of several independent -originally subjective - perspectives about music.

Message extraction from printed documents-a complete solution

The task to be solved within our core research was the design and development of a document analysis toolbox covering typical document analysis tasks such as document understanding, information extraction and text recognition. In order to prove the feasibility of our concepts, we have developed the prototypical analysis system OfficeMAID (Office Mail Analysis, Interpretation and Delivery). The system analyses documents, as used in the daily work of a purchasing department, by a priori knowledge about workflows and document features. In this way, the system provides goal-directed information extraction, shallow understanding and process identification for given documents (paper, fax, e-mail).

A simplified attributed graph grammar for high-level music recognition

This paper describes a simplified attributed programmed graph grammar to represent and process a-priori knowledge about common music notation. The presented approach serves as a high-level recognition stage and is interlocked to previous low-level recognition phases in our entire optical music recognition system (DOREMIDI++). The implemented grammar rules and control diagrams describe a declarative knowledge base to drive a transformation algorithm. This transformation converts the results of symbol recognition stages to a symbolic representation of the musical score.

A self-organizing map based knowledge discovery for music recommendation systems

In this paper, we present an approach for musical artist recommendation based on Self-Organizing Maps (SOMs) of artist reviews from Amazon web site. The Amazon reviews for the artists are obtained using the Amazon web service interface and stored in the form of textual documents that form the basis for the formation of the SOMs. The idea is to spatially organize these textual documents wherein similar documents are located nearby. We make an attempt to exploit the similarities between different artist reviews to provide insights into similar artists that can be used in a recommendation service. We introduce the concept of a modified weighting scheme for text mining in the musical domain and demonstrate its role in improving the quality of the recommendations. Finally, we present results for a list of around 400 musical artists and validate them using recommendations from a popular recommendation service.

Lexical postprocessing by heuristic search and automatic determination of the edit costs

We describe the realization of a dictionary based lexical postprocessing approach. A character hypotheses lattice (CHL) serves as input which is compared with the words of the vocabulary, using a generalization of the weighted edit distance. The search for the best word is based on a depth first traversal through the paths of the CHL and is directed by several heuristics to achieve a reasonable processing speed without deteriorating the recognition rate significantly. An iterative supervised automatic learning algorithm is proposed which determines the costs for the edit operations. Experiments reveal that this method significantly improves the recognition accuracy.

MUSIC SIMILARITY ANALYSIS IN A P2P ENVIRONMENT

In this paper we present our work towards an integrated P2P platform for convenient music retrieval. The main goal is to provide a distributed computation and storage of basic music features from MP3 encoded files and/or sources on the web. In this way we try to resolve the problem of missing meta data in order to provide intelligent services for convenient music querying. We believe that such value-based features are essential for a consumer-oriented success of future music distribution services.

Using natural language input and audio analysis for a human-oriented MIR system

We present a MIR (music information retrieval) system using natural language as input for human-oriented queries to large-scale music collections. The outlined system is a full-fledged architecture combining state-of-the-art approaches from the fields of natural language and the automatic analysis of audio data. Our approach copes with meta tag construction, content-based classification of audio and uses music ontologies as a backbone for the representation of musical knowledge. On top of this architecture different prototypes for industrial applications are described including first results of real-life field tests. This work has been performed at the German Research Center for AI and the authors spin-off company, the Sonicson GmbH.

Enhancing Music Recommendation Algorithms Using Cultural Metadata

In todays online commercial music marketplaces, a common requirement is to generate lists of artists that are “similar” to a given chosen artist. However, this is by no means a trivial task. A recent trend has been to tackle this challenge using sociocultural connotations rather than the traditional content-based audio or lyrics analysis. This article describes an enhancement to this approach that relies on the acquisition, filtering and condensing of unstructured, text-based information that can be found on the World Wide Web to recognize what the music community regards as “similar” artists. The beauty of this approach lies in its ability to access so-called “cultural metadata” (i.e., textual data about musical content) which is the aggregation of several independent – originally subjective – perspectives about a piece of music. The major focus of this work is the evaluation and enhancement of existing approaches in this area using filtering methods to increase their precision. A meaningful evaluation of the results is provided by a comparison with ground truth data.

A Journey to the Core of the Blogosphere

Blogs are popular communication instruments in todays web and altogether, they form the so-called blogosphere. This blogosphere has repeatedly been subject to structural analyses, and one of the findings has been the discovery of the A-List phenomenon, a cohesive group of influential blogs in the center of the blogosphere, whose exact identification remained an open issue. We use four language-specific subsets of the blogosphere, for which we aggregated the blogroll-based networks. We adapt core theory to analyse and compare the cohesion in these four data-sets, and provide a new scalable method for the identification of core-periphery structures in blog networks, which can contribute to identify A-List blogs more reliably.