Towards Meaningful Maps of Polish Case Law
نویسندگان
چکیده
In this work, we analyze the utility of two dimensional document maps for exploratory analysis of Polish case law. Such maps reflect the structure of analyzed collection by grouping similar documents in a neighbouring regions of 2D space. This visual aid could be useful for browsing and searching, finding anomalous documents or quickly gaining synthetic knowledge about large corpora. We started our study by selecting two automatic methods of generating document maps. First one is a standard method using linear principal component analysis (PCA) and visualizing two most important components on a scatterplot. Second one is a modern nonlinear method of embedding multidimensional data in 2D, namely, the t-Distributed Stochastic Neighbor Embedding method (t-SNE) [1]. t-SNE is optimized to preserve local structure (i.e., very similar documents should end up close together) and to reduce the so called ,,crowding problem”. The ,,crowding problem” is undesired feature of many dimensionality reduction techniques, which essentially place many data points in a very small area of low-dimensional space (hence creating a crowd). These two features make t-SNE unique and frequently used in recent analyses of high-dimensional data. The results of the PCA and t-SNE analysis of 2000 judgments from Polish courts are presented in Figure 1. t-SNE approach forms three well separated clusters which occupy large fraction of available space. For PCA, there is no obvious structure visible, as well as most of the available space remains unused. In Figure 2, we added color-coded information about the document source (which
منابع مشابه
The Role of Family in Abnormality and Crime of Children with a Case Study
Family is one of the most important grounds of crime and crime victimization development. The experts, welfare workers, lawyers and sociologists argue that the factor or factors should be identified that provide the kind of conditions under which the individuals abuse, tolerate or accept the crime in order to secure the society, reduce the social harms and finally propose the procedures to refo...
متن کاملScaffolding EFL Oral Performance through Story Maps and Podcasts and Students’ Attitudes toward it
The present study sought to explore the impact of story maps and audio podcasts as scaffolds on oral proficiency of Iranian EFL learners. The quasi-experimental study was launched with 36 EFL undergraduates in three groups by adopting a counterbalanced 3 3 Latin squared design. All participants were indiscriminately, but in a specified order, exposed to the three treatment conditions of story ...
متن کاملEconomic and environmental performance analysis of Polish energy companies
The tightening of the European Union climate and energy policy, directed to raising the resilience and effectiveness of the European Union Emissions Trading System, may influence on companies' economic performance. Polish energy companies, which mainly use coal in the energy production process, are particularly worried about the potential negative consequences associated with the implementation...
متن کاملSocial Factors Influencing Non-English Major Professors' Attitudes in Islamic Azad Universities of Third Area towards Learning English as an Academic Language (Case Study: Babol Branch)
Nowadays, professors of different universities around the world with different majors need to learn English for their academic advancement and research promotion. This requires that they have positive attitudes towards English language learning. In this regard, the role of social factors can be useful. Hence, this study was an endeavor to examine the social factors influencing non-English major...
متن کامل