Legal text and contexts are a rich domain to apply a range of NLP techniques. There are several factors that are fuelling developments. Large corpora of legal texts are increasingly publicly available, provided by different levels of governments around the world. Open linked data is a growing trend. The analytic frameworks with allied tools are well-developed. It is pressing to address the complex ramifications of laws, particularly in a global view, e.g. banking regulations. Government bodies have a strong agenda to reduce costs, increase efficiency, facilitate economic growth, provide transparent services, and engage with the public.
However, the law sets a very high bar on the requirements for the sorts and quality of analysis and the results of the applications of NLP tools, for detailed semantic content must be identified and extracted, the derived materials must be traced to the textual source, and the output ought to be well-justified if it is to serve as a basis for legally viable decision making. It is unclear that statistical NLP can alone meet these requirements. In addition, application of NLP to legal materials opens up novel domains of inquiry, e.g. network analysis of legislation and case law, semantic relations amongst cases, formalisation of legal concepts, and others.
The poster briefly outlines several aspects - conceptual problems, issues of corpora and scale, technical problems, use cases, and a sample of available technologies to address the materials and use cases.
Communications
The communication presentations (keynote talks and posters) will be continuously in May and June.
Dr. Wyner is a Lecturer at the University of Aberdeen, Scotland. His research interests are in Natural Language Processing, Semantics, and Argumentation, particularly as they bear on legal materials/purposes in legislation, regulations, case law, and policy-making.
Poster
Attachment | Size |
---|---|
CHIST-ERA Conference 2014 - Adam Wyner.pdf | 1.34 MB |
Dr. Adeline Lazarenko is Professor at Paris-Nord University. Her research interests include Natural Language Processing and Semantics.
Taking for granted that Human Language Understanding is a task-oriented process, we propose a data and knowledge based approach that combines textual sources, existing data and a minimum of human supervision to extract relevant information. This approach relies on machine reading and knowledge and data extraction. It enables a smooth integration of textual and world data and dynamically adapts to a given context and task. By simulating human-like ability to see only what is needed in a given context, it paves the way for many applications in e.g. text-based knowledge acquisition and data analytics. Research interests and/or link to main author's webpage: NLP, computational semantics, corpus semantics, corpus annotation, knowledge engineering, knowledge discovery, semantic web
Poster
Attachment | Size |
---|---|
CHIST-ERA Conference 2014 - Adeline Lazarenko.pdf | 1.19 MB |
Bilge Koroglu is Senior Software Developer at Yapi Kredi. Her research interests include Natural Language Processing and Machine Learning.
We have been receiving customer complaints from our branches, internet and telephone banking in digital format. These complaints are classified into predefined types of complaints reading by employees from customer services department of our bank. It requires extremely huge amount of human labelling process. We have been starting to work on a decision support system which classifies the complaints according to a statistical model so that human labelling is minimized. By using text categorization and machine learning algorithms, the system can learn a model dynamically and also can answer newly defined complaint class.
Poster
Attachment | Size |
---|---|
CHIST-ERA Conference 2014 - Bilge Koroglu.pdf | 179.33 KB |
Dr. Lavendelis is Assistant Professor at Riga Technical University in Latvia. The research interests of Dr. Lavendelis include Intelligent Agents, Multi-Agent Systems, Agent-Oriented Software Engineering, Intelligent Tutoring Systems, Ontology-Based Software Systems, Autonomous Software and Robotic Systems, Machine Learning.
The idea is to create a framework for autonomous research assistants (AREAs). We will design an infrastructure and protocols for machine-first (yet human-friendly) research communication, and create several AREA prototypes. While the most creative aspects of the research process must remain a human domain, AREAs will produce, exchange, consume and review research results; discover correlations in data generated by sensor networks and/or robots and pose hypotheses; conduct peer review of published research objects and assess reputation of researchers. In our interdisciplinary work, we will combine insights from the theory of multi-agent systems, scholarly communication studies, and text and data mining.
Poster
Attachment | Size |
---|---|
CHIST-ERA Conference 2014 - Egons Lavendelis.pdf | 316.05 KB |
Mr. Fatih Samet Cetin is Senior Software Development Specialist at Turkcell Global Bilgi in Turkey. As engineer, he works on Big Data analysis with ML Technologies, Smart Systems which can understand human language, IR applications for getting accurate data on accurate time.
Turkcell Global Bilgi is the leading customer relationship management center in Turkey. Social media has become one of customer contact channel recently. We have social media engagement, analytics and monitoring tool which is called Social Sniffer. In Social Sniffer we build valuable reports from social media for brands and companies.
Natural language usage in social media is the main challenge for improving the product’s text analytics features. Social media has its own slang language which consists of acronyms, emoticons, asciifications and so on. In addition to that, some of the interested languages have complex morphological structure and flexible constituent order. Therefore, special natural language processing tools are needed to work with this kind of languages.
We aim to present a bunch of advanced text analytics solutions for social media data, such as automated sentiment analysis, trending topic discovery, tagging and classifying social media posts, searching and filtering text contents. For improving our text analytics features, we are ready to work cooperatively.
Poster
Attachment | Size |
---|---|
CHIST-ERA Conference 2014 - Fatih Samet Cetin.pdf | 223.9 KB |
Dr. Gulsen Eryigit is an Assistant Professor at the Department of Computer Engineering of Computer and Informatics Faculty of Istanbul Technical University in Turkey. Her main research area is Natural Language Processing, and especially the application of Machine Learning techniques into this area.
Despite the importance of structural analysis in properly determining sentence meaning, many commercial Text Analytics solutions do not employ parsing. One reason for this is that it is only very recently that automatic parsers have become efficient enough to be used as part of a content processing workflow operating on web-scale data sets. While the past fifteen years has seen significant advances in statistical parsing, two crucial weaknesses remain:
1. Multilingualism
2. Noisy Web 2.0 Text
The Parse4Real project will explicitly address these two challenges by producing multilingual, adaptive and scalable parsing technology for Web2.0 text.
Poster
Attachment | Size |
---|---|
CHIST-ERA Conference 2014 - Gulsen Eryigit.pdf | 381.69 KB |
Dr. Holger Scwenk is Professor at the University of Le Mans, in France. His research interests include Machine Learning, Deep Neural Networks, Statistical Machine Translation.
We propose to work on the application of novel machine learning algorithms, in particular deep neural networks, to achieve significant breakthroughs in the field of human language understanding. We will work on speech recognition, statistical machine translation, dialog systems and computer vision. All these tasks will be jointly addressed. To support these activities we will also create and distribute the appropriate multimodal corpora.
Poster
Attachment | Size |
---|---|
CHIST-ERA Conference 2014 - Holger Schwenk.pdf | 201.19 KB |
Dr. Iryna Sekret is Professor at the Zirve University in Turkey. Her research interests are in the areas of psychology and language, namely educational psychology, ICT and e-learning, social media in learning, contrastive linguistics, semantics and translation, English language teaching, and teaching English for specific purposes.
The research proposal is within the area of the semantic analysis to improve quality of e–translation and language e–learning. Among urgent problems are:
1) low quality of the machine translation because of the scarcity of efficient language solutions based on the thorough semantic analysis of the language means regarding their derivation, compatibility with other linguistic units, processes of the semantic shift occurring in cases of the language unit borrowing, its dialectal usage, or representation in the jargon speech;
2) lack of e-learning programmes for developing effective vocabulary skills;
3) problems of the semantic analysis of the sign language.
Poster
Attachment | Size |
---|---|
CHIST-ERA Conference 2014 - Iryna Sekret.pdf | 270.51 KB |
- 1 of 2
- next ›