Improved Recognition and Normalisation of Polish Temporal Expressions

Kocoń, Jan; Marcińczuk, Michał

doi:10.26615/978-954-452-049-6_051

Cited by 4 publications

(4 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We used WCRFT tagger [29], which utilises Toki [30] to tokenise the input text before the creation of the embeddings model. The comparison of EC1 with previous results obtained using only CRF [9] show the significant improvement across all the tested metrics: 3.6pp increase in strict F1-score, 1.36pp increase in relaxed precision, 5.61pp increase in relaxed recall and 3.51pp increase in relaxed F1-score. Table 9: Evaluation results for all TIMEX3 classes (total) for 9 word embeddings models (3 best models from each embeddings group: EE, EP, EC from Table 8) using the following measures from [35]: strict precision, strict recall, strict F1-score, relaxed precision, relaxed recall, relaxed F1-score, type F1-score.…”

Section: Discussionmentioning

confidence: 66%

“…We trained the final models using the train set and we evaluated it using the test set, which was the reproduction of analysis performed in articles [11,9]. The division is presented in Table ??.…”

Section: Experiments and Resultsmentioning

confidence: 99%

“…The best systems listed in [35], called TIPSem [16] and ClearTK [1], use CRFs for recognition, so initially, we decided to apply the CRF-based approach for this task. The results were described in [12,10].…”

Section: Recognition Of Timexesmentioning

confidence: 99%

See 2 more Smart Citations

Evaluating KGR10 Polish word embeddings in the recognition of temporal expressions using BiLSTM-CRF

Kocoń¹,

Gawor²

2019

Preprint

Self Cite

View full text Add to dashboard Cite

The article introduces a new set of Polish word embeddings, built using KGR10 corpus, which contains more than 4 billion words. These embeddings are evaluated in the problem of recognition of temporal expressions (timexes) for the Polish language. We described the process of KGR10 corpus creation and a new approach to the recognition problem using Bidirectional Long-Short Term Memory (BiLSTM) network with additional CRF layer, where specific embeddings are essential. We presented experiments and conclusions drawn from them.

show abstract

Section: Discussionmentioning

confidence: 66%

Section: Experiments and Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Evaluating KGR10 Polish word embeddings in the recognition of temporal expressions using BiLSTM-CRF

Kocoń¹,

Gawor²

2019

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…This tool was successfully used in other Natural Language Engineering tasks, mainly in Named Entities Recognition Marcińczuk and Kocoń (2013); Marcińczuk et al (2013). We described our first approach to recognise timexes using this tool Kocoń and Marcińczuk (2015) and this work extends that research.…”

Section: Recognitionmentioning

confidence: 57%

Supervised approach to recognise Polish temporal expressions and rule-based interpretation of timexes

Kocoń¹,

Marcińczuk²

2016

Nat. Lang. Eng.

Self Cite

View full text Add to dashboard Cite

A key challenge of the Information Extraction in Natural Language Processing is the ability to recognise and classify temporal expressions (timexes). It is a crucial source of information about when something happens, how often something occurs or how long something lasts. Timexes extracted automatically from text, play a major role in many Information Extraction systems, such as question answering or event recognition. We prepared a broad specification of Polish timexes – PLIMEX. It is based on the state-of-the-art annotation guidelines for English, mainly TIMEX2 and TIMEX3 (a part of TimeML – Markup Language for Temporal and Event Expressions). We have expanded our specification for a description of the local meaning of timexes, based on LTIMEX annotation guidelines for English. Temporal description supports further event identification and extends event description model, focussing on anchoring events in time, events ordering and reasoning about the persistence of events. We prepared the specification, which is designed to address these issues, and we annotated all documents in Polish Corpus of Wroclaw University of Technology (KPWr) using our annotation guidelines. We also adapted our Liner2 machine learning system to recognise Polish timexes and we propose two-phase method to select a subset of features for Conditional Random Fields sequence labelling method. This article presents the whole process of corpus annotation, evaluation of inter-annotator agreement, extending Liner2 system with new features and evaluation of the recognition models before and after feature selection with the analysis of statistical significance of differences. Liner2 with presented models is available as open source software under the GNU General Public License.

show abstract

Multi-task Sequence Classification for Disjoint Tasks in Low-resource Languages

Radom

Kocoń

2021

Procedia Computer Science

View full text Add to dashboard Cite

Improved Recognition and Normalisation of Polish Temporal Expressions

Cited by 4 publications

References 15 publications

Evaluating KGR10 Polish word embeddings in the recognition of temporal expressions using BiLSTM-CRF

Evaluating KGR10 Polish word embeddings in the recognition of temporal expressions using BiLSTM-CRF

Supervised approach to recognise Polish temporal expressions and rule-based interpretation of timexes

Multi-task Sequence Classification for Disjoint Tasks in Low-resource Languages

Contact Info

Product

Resources

About