Tamás Váradi

Towards Linked Language Data for Digital Humanities

We investigate the extension of classification schemes in the Humanities into semantic data repositories, the benefits of which could be the automation of so far manually conducted processes, such as detecting motifs in folktale texts. In parallel, we propose linguistic analysis of the textual labels used in these repositories. The resulting resource, which we propose to publish in the Linked Open Data (LOD) framework, will explicitly interlink domain knowledge and linguistically enriched language data, which can be used for knowledge-driven content analysis of literary works.

Using Deep Rectifier Neural Nets and Probabilistic Sampling for Topical Unit Classification

In the interaction between humans and computers as well as in the interaction among humans, topical units (TUs) have an important role. This motivated our investigation of topical unit recognition. To lay foundations for this, we first create a classifier for topical units using Deep Neural Nets with rectifier units (DRNs) and the probabilistic sampling method. Evaluating the resulting models on the HuComTech corpus using the Unweighted Average Recall (UAR) measure, we find that this method produces significantly higher classification scores than those that can be achieved using Support Vector Machines, and what DRNs can produce in the absence of probabilistic sampling. We also examine experimentally the number of topical unit labels to be used. We demonstrate that not having to discriminate between variations of topic change leads to better classification scores. However, there can be applications where this distinction is necessary, for which case we introduce a hierarchical classification method. Results show that this method increases the UAR scores by more than 7%.

Beyond Sentiment: Social Psychological Analysis of Political Facebook Comments in Hungary

This paper presents the methodology and results of a project for the large-scale analysis of public messages in political discourse on Facebook, the dominant social media site in Hungary. We propose several novel social psychology- motivated dimensions for natural language processing-based text analysis that go beyond the standard sentiment-based analysis approaches. Communion describes the moral and emotional aspects of an individual’s relations to others, while agency describes individuals in terms of the efficiency of their goal- orientated behavior. We treat these by custom lexicons that identify positive and negative cues in text. We measure the level of optimism in messages by examining the ratio of events talked about in the past, present and future by looking at verb tenses and temporal expressions. For assessing the level of individualism, we build on research that correlates it to pronoun dropping. We also present re- sults that demonstrate the viability of our measures on 1.9 million downloaded public Facebook comments by examining correlation to party preferences in public opinion poll data.

Detecting Gaps in Language Resources and Tools in the Project CESAR

In this paper the first preliminary results of the analysis of marks collected within the tables of META-NET series of Language White Papers of CESAR project languages are demonstrated. Although they are preliminary results, we can consider them useful for showing us where real gaps in language resources and tools can be detected.

Rule-based partial MT using enhanced finite-state grammars in NooJ

The paper argues for the viability and utility of partial machine translation (MT) in multilingual information systems. The notion of partial MT is modelled on partial parsing and involves a bottomup pattern matching approach where the finite-state transducers assign translation equivalents locally. The article focuses on the linguistic underpinnings of the approach and gives illustrations of its implementation within the NooJ finite-state linguistic development system.