Werner Winiwarter

Adaptive natural language interfaces to FAQ knowledge bases

Abstract In this paper, we present a natural language interface architecture, which is used for accessing FAQ knowledge bases. Since one of the main obstacles to the efficient use of natural language interfaces is the high amount of required manual knowledge engineering, we provide an adaptive architecture to automate the acquisition of linguistic knowledge. We apply a machine learning module based on an eXtended Semantic Enumeration (XSE) tree to activate linguistic tests from semantic, syntactic, and pragmatic analyses during the traversal of the tree. The XSE-tree is built automatically based on past questions without the need for any additional linguistic knowledge.

Atmospheric concentrations of formic and acetic acid and related compounds in eastern and northern Austria

Abstract Formic and acetic acid were measured in the gas phase at three sites in eastern and northern Austria using an annular diffusion denuder sampling technique. The daytime background concentration of formic acid averaged 1.4 ±0.2 ppb (63 ± 9 nmol m−3) in a semirural area in eastern Austria. In a rural area in Austria the corresponding level was 0.9 ±0.3 ppb (40 ± 13 nmol m−3). About twice as high levels of formic acid were found during a photochemically active summer period and a wintertime pollution episode. The concentration of acetic acid was generally about 50 % lower than that of formic acid during daylight hours. In summer diurnal variation with a night-time minimum and a daytime maximum in the late afternoon hours was observed. The diurnal variation was not evident in results from measurements above the inversion layer or during winter conditions with snow cover. No increase in the formic and acetic acid levels was found in the airshed of Vienna, indicating that traffic emissions do not form a major primary source for the semirural concentrations measured. Although there might be biogenic sources for formic and acetic acid, our findings suggest a noticeable contribution from various anthropogenic emissions being a source for photochemical production of the acids.

Determination of inorganic and organic volatile acids, NH3, particulate SO 4 2? , NO 3 2? and Cl? in ambient air with an annular diffusion denuder system@@@Bestimmung der atmosphrischen Konzentrationen von anorganischen und organischen gasfrmigen Suren, NH3, teilchenfrmigem SO 4 2? , NO 3 2? und Cl? mit einem Ringspaltrohr-Diffusionsdenudersystem

SummaryAn annular denuder system for simultaneous determination of gaseous and particulate pollutants in ambient air was developed. Inorganic acids (HCl, HNO3) were collected in a NaF-coated denuder, while organic acids (HCOOH, CH3COOH) were trapped in one coated with KOH. NH3 was sampled with a H3PO4-coated denuder tube. Particulate H2SO4 was evaporated at elevated temperature (145°–155°C) and deposited on a NaF-coating together with HCl and HNO3 originating from thermal decomposition of NH4Cl and NH4NO3. NH3 resulting from deammoniation of (NH4)2SO4 as well as NH3 remaining from NH4Cl and NH4NO3 were collected in a H3PO4-coated denuder. The practical collection capacity of the tubes ranged from 0.35 to 1.0 μmol calculated from an experimentally determined sorption efficiency of at least 90%. The precision, expressed as relative standard deviation, of sampling and analytical procedures was determined from duplicate measurements in ambient air. The reproducibility varied from 9% to 14% for the gaseous components, while that of the particulate compounds ranged from 12% to 23%. Aqueous extracts of the denuder coatings were analysed for ionic components by ion chromatography using conductivity detection. The minimum detectable concentration in air was found to be 1.5 to 14 nmol/m3 for the different compounds calculated for 1.6 m3 sample volume, based on 3-h measurements at a flow of 9 l/min.

Modelling data secrecy and integrity

The paper describes a semantic data model used as a design environment for multilevel secure database applications. The proposed technique is built around the concept of security classification constraints (security semantics) and takes into account that security restrictions may either have effects on the static part of a system, on the behavior of the system (the system functions), or on both. As security constraints may influence each other appropriate integrity mechanisms are necessary and modelling of a multilevel application must be data as well as function driven. This functionality is included in the proposed semantic data model for multilevel security by developing secure data schemas, secure function schemas, a procedure for alternating iterative refinements on either schema, and a powerful integrity system to check the consistency of the classification constraints and of the multilevel secure database application.

Using advanced transaction meta‐models for creating transaction‐aware web service environments

Recently, the software industry has published several proposals for transactional processing in the Web service world. Even though most proposals support arbitrary transaction models, there exists no standardized way to describe such models. This paper describes potential impacts and use cases of utilizing advanced transaction meta‐models in the Web service world and introduces two suitable meta‐models for defining arbitrary advanced transaction models. In order to make these meta‐models more usable in Web service environments, they had to be enhanced, and XML representations of the enhanced models had to be developed.

The Entity-Relationship Model for Multilevel Security

A design environment for security critical database applications that should be implemented by using multilevel technology is proposed. For this purpose, the Entity-Relationship model is extended to capture security semantics. Important security semantics are defined and a language to express them in an ER model by means of security constraints is developed. The main contribution consists of the development and implementation of a rule-based system with which security semantics specified may be checked for conflicting constraints. The check involves application independent as well as application dependent integrity constraints and leads to a non conflicting conceptual representation of the security semantics of a multilevel secure database application.

CONCAT - Connotation Analysis of Thesauri Based on the Interpretation of Context Meaning

Knowledge acquisition constitutes the bottleneck for the creation of legal expert systems. Legal language must be formalised to such a degree that it can be processed automatically. We deal with this problem by supporting the process of creating a selective thesaurus for a legal information system which can be seen as prerequisite for further knowledge processing. This selectivity is obtained by means of connotation analysis of the individual descriptors which makes it possible to detect hidden word meanings and to distinguish between precise legal terms and words with fuzzy meaning. Within the prototype system CONCAT we applied both a statistical and a connectionist approach to connotation analysis and performed a comparative evaluation of the achieved results.

Exploiting syntactic analysis of queries for information retrieval

Up to now, the results of applying sophisticated NL techniques to information retrieval (IR) have been mostly disappointing. Our research aims at investigating in detail the role of syntactic analysis in IR and at finding answers to the question why it works better for some queries and worse for others. The final goal is a hybrid algorithm that selectively applies syntactic analysis to certain classes of queries while relying on standard statistical techniques otherwise.

Information filtering: the computation of similarities in large corpora of legal texts

TraditiottaJ information retrieval systems do not satis& the lawyers’ demands because they provide only syntactic representation of legal data, The bottleneck for the creation of the more promising conceptual information retrieval systems is the time-consuming knowledge acquisition. The best solution is the representation of legal knowledge by simple linguistic tools, statistics and neural networks. In our prototype KONTERM we represent legal knowledge about concepts and documents by a knowledge base which is structured by statistical and connectionist methods. In fitture, this knowledge base will be used to filter legal knowledge from documents.

Natural language processing technologies for developing a language learning environment

So far, Computer-Assisted Language Learning (CALL) comes in many different flavors. Our research work focuses on developing an integrated e-learning environment that allows improving language skills in specific contexts. Integrated e-learning environment means that it is a Web-based solution that performs language learning tasks using common working environments like, for instance, Web browsers or Email clients. It should be accessible on different platforms, even on mobile devices. Natural Language Processing (NLP) forms the technological basis for developing such a learning framework. The paper gives an overview of the state-of-the-art in this area. Therefore, on the one hand, it explains creation processes for NLP resources and gives an overview of corpora. On the other hand, it describes existing NLP standards. Based on our requirements, the paper gives special attention to the evaluation and comparison of toolkits that can suitably support the planned implementation. An outlook at the end points out necessary developments in e-learning to keep in mind.