Günther Fliedl

Classification of customer reviews based on sentiment analysis.

In this paper we propose a system that performs the classification of customer reviews of hotels by means of a sentiment analysis. We elaborate on a process to extract a domainspecific lexicon of semantically relevant words based on a given corpus (Scharl et al., 2003; Pak & Paroubek, 2010). The resulting lexicon backs the sentiment analysis for generating a classification of the reviews. The evaluation of the classification on test data shows that the proposed system performs better compared to a predefined baseline: if a customer review is classified as good or bad the classification is correct with a probability of about 90%.

From textual scenarios to a conceptual schema

Scenarios are a very popular means for describing and analyzing behavioral aspects on the level of natural language. In information systems design, they form the basis for a subsequent step of conceptual dynamic modeling. To enhance this step, linguistic instruments prove applicable for transforming scenarios into conceptual schemas of various models. This transformation usually consists of three steps: linguistic analysis, component mapping and schema construction. Within this paper we investigate to which extent these steps may be performed automatically in the framework of KCPM, a conceptual predesign model which is used as an Interlingua between natural language and arbitrary conceptual models.

Linguistically based requirements engineering — the NIBA-project

Abstract Usually, the development of an information system (or some part of it) starts with requirement elicitation followed by a phase of collecting and analyzing which results in a set of requirements specifications. As opposed to conventional conceptual modeling, where input texts are formalized, our approach suggests the idea of collecting and cataloguing natural language patterns in a non-textual form immediately after a linguistic analysis. This linguistic analysis is done according to the NTMS model. Collecting and cataloguing of natural language data is supported by KCPM.

From OWL class and property labels to human understandable natural language

The ontology language OWL has become increasingly important during the previous years. However due to the uncontrolled growth, OWL ontologies in many cases are very heterogeneous with respect to the class and property labels that often lack a common and systematic view. For this reason we linguistically analyzed OWL class and property labels focusing on their implicit structure. Based on the results of this analysis we generated a first proposal for linguistically determined label generation which can be seen as a prerequisite for mapping OWL concepts to natural language patterns.

Guideline based evaluation and verbalization of OWL class and property labels

The ontology language OWL has become increasingly important during the previous years. However due to the uncontrolled growth, OWL ontologies in many cases are very heterogeneous with respect to the class and property labels that often lack a common and systematic view. For this reason we developed linguistically based guidelines for OWL class and property labels focusing on their implicit structure. Considering these guidelines we propose an evaluation mechanism including rules for comparing the linguistically triggered label interpretations to their OWL internal representations. Our proposal also includes the verbalization of these evaluated OWL labels.

Text Preparation Through Extended Tokenization

Tokenization is commonly understood as the first step of any kind of natural language text preparation. The major goal of this early (pre-linguistic) task is to convert a stream of characters into a stream of processing units called tokens. Beyond the text mining community this job is taken for granted. Commonly it is seen as an already solved problem comprising the identification of word borders and punctuation marks separated by spaces and line breaks. But in our sense it should manage language related word dependencies, incorporate domain specific knowledge, and handle morphosyntactically relevant linguistic specificities. Therefore, we propose rulebased Extended Tokenization including all sorts of linguistic knowledge (e.g., grammar rules, dictionaries). The core features of our implementation are identification and disambiguation of all kinds of linguistic markers, detection and expansion of abbreviations, treatment of special formats, and typing of tokens including singleand multi-tokens. To improve the quality of text mining we suggest linguistically-based tokenization as a necessary step preceeding further text processing tasks. In this paper, we focus on the task of improving the quality of standard tagging.

\“NIBA – TAG” - A Tool For Analyzing And Preparing German Texts

NIBA-TAG is some kind of multilevel natural language tagger with rich functionality. It functions as a word-stemmer, a morphological parser and a normal POS-Tagger, which uses syntactic und semantic features for contextually influenced word-tagging. Each rule is based on a ranking-mechanism which is currently related to the levels “fact”, “proposal” and “guess”. One of the postprocessing-units analyzes the ranking-structure and can change a “proposal” to a “fact”, if enough rules made an identical proposal for a word. The default output is XML, where the level of precision can be specified. So one could generate a XML-file only including the guesses, or a file with all attributes relevant for the status of a proposal.

Semantic Tagging and Chunk-Parsing in Dynamic Modeling

The paper outlines a multilevel tagging approach to the linguistic analysis of requirements texts. It is shown that extended semantic tagging including chunk-parsing of noun-groups and prepositional groups enables to identify structural items which can be mapped to the conceptual notions for dynamic modeling in KCPM, namely actor, operation-type and condition.

Extended tagging and interpretation tools for mapping requirements texts to conceptual (predesign) models

This paper discusses an advanced tagging concept which supplies information that allows for a tool supported step by step mapping of natural language requirements specification to a conceptual (predesign) model.The focus lies on sentences containing conditions as used for describing alternatives in business process specifications. It is shown, how the tagging results are interpreted systematically such allowing for a stepwise model generation.

Guidelines for NL-Based Requirements Specifications in NIBA

In this paper we discuss the linguistic base of standardized sentences as well as their structure and their employment in requirements analysis. Standardized sentences are quite helpful in the automatic analysis of requirements concerning static and dynamic aspects. Since standardized sentences require being filtered out of prose texts, which is a time consuming task we prefer to use standardized sentences in the case of requirements completion. Such completions inevitably emerge in the iterative process of requirements analysis. To enhance the process of requirements analysis we use a special approach called conceptual predesign the results of which are mapped by heuristic rules to conceptual design schemes, e.g. formulated in UML.