 |
Chair of Computer Science 8: Artificial Intelligence
The AI department was founded in 1991. Research and teaching is
centered around the fields
Knowledge Representation, Natural Language Processing, Mobile AI,
Symbolic Scientific Computing, and Digital Documents.
Furthermore, we offer lectures on the Foundations of Logic,
Media Informatics, and Software Development for the Internet.
- Natural language processing concerns algorithms which are able to analyze and generate both written and spoken language, as well as applying the findings to natural language based human-computer interaction.
- Text mining includes classification, retrieval and grouping of documents, as well as extracting parts of interest within these. These basic techniques are used in several of our research projects. Our findings on semantic abstraction i.e. the extraction of topics or other features from documents are particularly noteworthy.
- As an area within NLP, computational semantics aims at constructing formal representations of the meaning of phrases and interpreting them. Following syntactical relations between words and phrases, the meanings of single words are put together forming hypotheses on the semantics of the phrase as a whole. At the AI department, discourse representation theory is being extended to capture semantics in human-computer interaction.
- In IGSuS, we develop new means for organizing TV channels to enable convenient browsing even in the light of several hundred channels.
- ROSE is mobile navigation with support of public transport. It incorporates personal navigation, position and time dependent search for transport links and web 2.0 technology. ROSE makes it easier for passengers to use public transport systems. It determines the best possible transport link and accompanies passengers throughout their entire journey. Our motivation is to free the passenger from as many tedious tasks as possible.
The user does not even have to know the destination! This is because ROSE will automatically propose destinations matching the interests of the user. For example, the user could ask if there are any outdoor music events today and ROSE will search huge databases, provided by our partners for available events and propose them to the user. The user can select an event and ROSE will determine where the event takes place and calculate how to best reach it using public transport. Users can also rate the proposals the system provides. Over time ROSE will get to know the users interests more accurately and provide even better proposals the next time.
- High word error rates for speaker-independent speech recognition present a performance barrier for modern dialog systems. This highly constrains the inputted speech both syntactically and lexically and means that it is not possible for the user to speak in the intuitive and spontaneous way that is characteristic of normal human conversation. This soon leads to frustration and negatively influences the user's perception of the system as a whole. The research project |Sprachfehler| combines methods from parsing, computational semantics and text mining with an empirical error model, aiming at identifying speech recognition errors and reconstructing the original utterance.
- This project is embedded in the context of the INI.FAU cooperation between the University of Erlangen and the German car manufacturer AUDI AG. Our concrete project partner is the sub-department for diagnosis systems of the electronic development department of AUDI AG.The project aims at the investigation of formal methods (meta-models, representations, algorithms) as a basis for tool suppport of the main tasks of the department, namely the development of the self-diagnosis functionalities (Onboard Diagnosis) of electronic control units (ECUs), which is achieved in a complex distributed manner with numerous different suppliers.Besides the verification of interface definitions enabling the data exchange between ECUs and offboard diagnosis systems, a second, for both sides fruitful area, currently subject to intense world-wide research efforts, has been identified: The specification-based generation of black-box tests for the fault detection functionalities of ECUs, modelled as appropriate reactive real-time systems.
- The project "WissKI - Infrastructure for Science Communication", funded by the
- The cosmographic and geographic tradition of the late middle ages (13th to 16th century) aims at orientation in the sense of a world view ('Weltanschauung'). The mappaemundi, which are an offspring of Christian encyclopedism, are first of all cognitive maps. In the center of the project there is the construction of a database which collects representative medieval and early modern maps of the world (mappaemundi, Ptolemaic maps and portulans), which will serve as a basis for the elaboration of the cognitive relations represented in them as well as their change.
The mappae database has been designed as a research tool; therefore the compilation of high resolution digital images and associated metadata has been emphasized from the very beginning.
|
 |