Dilek Küçük

Named Entity Recognition Experiments on Turkish Texts

Named entity recognition (NER) is one of the main information extraction tasks and research on NER from Turkish texts is known to be rare. In this study, we present a rule-based NER system for Turkish which employs a set of lexical resources and pattern bases for the extraction of named entities including the names of people, locations, organizations together with time/date and money/percentage expressions. The domain of the system is news texts and it does not utilize important clues of capitalization and punctuation since they may be missing in texts obtained from the Web or the output of automatic speech recognition tools. The evaluation of the system is performed on news texts along with other genres encompassing child stories and historical texts, but as expected in case of manually engineered rule-based systems, it suffers from performance degradation on these latter genres of texts since they are distinct from the target domain of news texts. Furthermore, the system is evaluated on transcriptions of news videos leading to satisfactory results which is an important step towards the employment of NER during automatic semantic annotation of videos in Turkish. The current study is significant for its being the first rule-based approach to the NER task on Turkish texts with its evaluation on diverse text types.

Enhanced Nationwide Wind-Electric Power Monitoring and Forecast System

This paper describes an enhanced monitoring and forecast system for the electrical power generated from wind in Turkey. Wind power plant (WPP) owners, transmission system operator, distribution system operators, and renewable energy experts are the shareholders of this system. The developed monitoring and forecast system is composed of wind electricity analyzers and reference wind masts, one for each WPP, a Wind-Electric Power Monitoring and Forecast Center (WPMFC) equipped with database, data processing, and application servers, and the associated monitoring and forecast software. All shareholders can communicate with the developed WPMFC in order to retrieve preprocessed wind-electric power forecast data and power quality data, through the IP network according to their authorization level. The electrical and meteorological raw data can also be retrieved for postprocessing. The developed system is based on global positioning system (GPS) synchronized, real-time field measurements taken from each WPP, in addition to the use of mesoscale numerical weather prediction models. Instantaneous electrical and meteorological quantities of each WPP can be monitored continuously through the WPMFC. The forecast system employs physical, statistical, and hybrid models combining both approaches. Some sample monitoring and forecast results are given in the paper to illustrate the benefits and abilities of the developed system.

A Hybrid Named Entity Recognizer for Turkish with Applications to Different Text Genres

In this study, we present a hybrid named entity recognizer for Turkish, which is based on a previously proposed rule based recognizer. Since rule based systems for speci¯c domains require their knowledge sources to be manually revised when ported to other domains, we turn the rule based recognizer into a hybrid one so that it learns from annotated data and improves its knowledge sources accordingly. Both the hybrid recognizer and its predecessor are evaluated on the same corpora and the hybrid recognizer achieves comparably better results. The current study is signi¯cant since it presents the¯rst hybrid {manually engineered and learning{ named entity recognizer for Turkish texts

Employing named entities for semantic retrieval of news videos in Turkish

Named entities are known to be important means for semantic annotation of news texts. Considerable work has been carried out for semantic indexing of both textual news and news videos especially in English through the employment of named entities extracted from textual news or transcriptions of the news videos. In this paper, we present our semantic retrieval architecture for news videos in Turkish based on prior semantic annotation of the videos with the corresponding named entities in the news transcription texts. We employ a rule-based named entity recognizer for Turkish which makes use of handcrafted sets of lexical resources and pattern bases. We compiled a small corpus of Turkish news videos and the named entity recognizer in its current form achieves a success rate of about 75% on this corpus. A retrieval interface is implemented to access the video corpus through the boolean queries formed with the extracted named entities. The interface currently does not involve any ranking procedure, displaying all the videos, the transcription texts of which satisfy the boolean query posed through the interface, sorted by their broadcast date. The presented study is significant for its being the first study to perform automatic semantic video annotation on a genuine news video corpus in Turkish and demonstrating the utilization of the annotations through a retrieval interface.

Building an ontology for flexible power quality querying

Electrical power quality (PQ) is measured as a set of PQ parameters, such as frequency, flicker, and harmonics, obtained from electrical transmission and distribution systems. The domain of electrical PQ suffers from the lack of a common vocabulary to be shared by the involved parties. There exist several systems of varying complexities to measure PQ parameters, yet no ontology development effort for PQ data has been reported in the literature. In this paper, a novel and generic domain ontology for electrical PQ is presented. During the ontology development process, relevant standards as well as experiences gained during the development of two distinct large-scale systems for nationwide PQ monitoring have been used. Among these monitoring systems, one is a mobile measurement system which has collected sample PQ data of considerable size all over the Turkish electricity transmission system to detect problematic locations. The second system, which aims to perform on line PQ monitoring with permanent monitors, is the successor of the previous one and has recently started to collect data at several problematic points in the transmission system. The PQ ontology is especially important for its being the first attempt to create a common vocabulary for the domain and hence to make differing PQ monitoring systems interoperable. The proposed ontology is utilized to develop a natural language based interface for querying the PQ data collected through mobile measurement system, in a flexible way. The implementation details of the interface is presented with a query example on the PQ data. The future work will include the extension of this natural language based interface into a speech recognition engine to get spoken queries into the system and enhance the interface to handle queries in Turkish.

Multilingual video indexing and retrieval employing an information extraction tool for turkish news texts: a case study

In this paper, a multilingual video indexing and retrieval system is proposed which relies on an information extraction tool, a hybrid named entity recognizer, for Turkish to determine the semantic annotations for the considered videos. The system is executed on a set of news videos in English and encompasses several other components including an automatic speech recognition system for English, an English-to-Turkish machine translation system, a news video database, and a semantic video retrieval interface. The performance evaluation demonstrates that the system components achieve promising results which provides evidence for the applicability of the system. The proposed system and its application on the video set are significant as they constitute a plausible case study targeting at the problem of multilingual video indexing and retrieval utilizing information extraction as the central technique for semantic video indexing.

A named entity recognition dataset for Turkish

Named entity recognition is one of the important topics in the research area of natural language processing. Named entity recognition studies conducted on Turkish texts are quite limited, compared to the studies on other languages. Besides, the lack of common data sets makes the comparison of different approaches harder. In this study, a dataset comprising news articles in Turkish annotated with named entities is presented. The annotations comprise the basic named entity types of person, location, and organization names. Additionally, to be used as reference in future studies, a rule-based named entity recognition system is evaluated on the final form of this data set and the corresponding evaluation results are presented. It is envisioned that our study will contribute to the advancement of named entity recognition studies on Turkish texts.

Improving automatic semantic annotations of news videos in Turkish through Web alignment and event extraction

Video texts — if available — constitute a valuable source for automatic semantic annotation of large video archives. In this paper, we present our attempts towards the improvement of a text-based semantic annotation and retrieval system for Turkish news videos through automatic Web alignment and event extraction. The results of our initial experiments turn out to be promising and these two features are incorporated into the existing system. Although the ideas of automatic Web alignment and text-based event extraction are not the novel contributions of the current paper, to the best of our knowledge, their first implementation and employment in a system for Turkish news videos is a significant contribution to related work on videos in lesser studied languages such as Turkish. Also overviewed in the current paper is the prospective version of the system encompassing components for several other tasks including topic segmentation, keyphrase extraction, news categorization and summarization to enhance the overall system.