Past prescriptions constitute a central element in patient records. These are often written in an unstructured and brief form. Extracting information from such prescriptions enables the development of automated processes in the medical data mining field. This paper presents a Conditional Random Fields (CRFs) based approach to extract relevant information from prescriptions. We focus on Finnish language prescriptions and make use of Finnish language specific features. Our labeling accuracy is 95%, which compares favorably to the current state-of-the-art in English language prescriptions. This, to the best of our knowledge, is the first such work for the Finnish language.

MorAz is an open-source morphological analyzer for Azerbaijani Turkish. The analyzer is available through both as a website for interactive exploration and as a RESTful web service for integration into a natural language processing pipeline. MorAz implements the morphology of Azerbaijani Turkish following a two-level approach using Helsinki finite-state transducer and wraps the analyzer with python scripts in a Django instance.

In this article, we summarize the methodology and the results of our 2-year-long efforts to construct a comprehensive WordNet for Turkish. In our approach, we mine a dictionary for synonym candidate pairs and manually mark the senses in which the candidates are synonymous. We marked every pair twice by different human annotators. We derive the synsets by finding the connected components of the graph whose edges are synonym senses. We also mined Turkish Wikipedia for hypernym relations among the senses. We analyzed the resulting WordNet to highlight the difficulties brought about by the dictionary construction methods of lexicographers. After splitting the unusually large synsets, we used random walk–based clustering that resulted in a Zipfian distribution of synset sizes. We compared our results to BalkaNet and automatic thesaurus construction methods using variation of information metric. Our Turkish WordNet is available online.

Large scale dictionaries such as wordnets are one of the important structures used in natural language processing. Wordnet is a comprehensive dictionary containing semantic relations including synonyms, antonyms, etc. In this paper, we try to extract Hypernym-Hyponym relations which are one of the main parts of WordNet. For this aim, we used Wikipedia, Contemporary Dictionary of Turkish and Wiktionary as corpora. We used Finite State Machines for developing several rules for Hypernym-Hyponym extraction.

We give a FST description of nominal and finite verb morphology of Azarbaijani Turkish.We use a hybrid approach where nominal inflection is expressed as a slot-based paradigm and major parts of verb inflection are expressed as optional paths on the FST.We collapse adjective and noun categories in a single nominal category as they behave similarly as far as their paradigms are concerned.Thus, we defer a more precise identification of POS to further down the NLP pipeline.

In this paper, we report our tree based statistical translation study from English to Turkish. We describe our data generation process and report the initial results of tree-based translation under a simple model. For corpus construction, we used the Penn Treebank in the English side. We manually translated about 5K trees from English to Turkish under grammar constraints with adaptations to accommodate the agglutinative nature of Turkish morphology. We used a permutation model for subtrees together with a word to word mapping. We report BLEU scores under simple choices of inference algorithms.

In this paper, we describe our initial efforts for creating a Turkish constituency parse treebank by utilizing the English Penn Treebank. We employ a semi-automated approach for annotation. In our previous work [18], the English parse trees were manually translated to Turkish. In this paper, the words are semi-automatically annotated morphologically. As a second step, a rule-based approach is used for refining the parse trees based on the morphological analyses of the words. We generated Turkish phrase structure trees for 5143 sentences from Penn Treebank that contain fewer than 15 tokens. The annotated corpus can be used in statistical natural language processing studies for developing tools such as constituency parsers and statistical machine translation systems for Turkish.

In this paper, we report our work on chunking in Turkish. We used the data that we generated by manually translating a subset of the Penn Treebank. We exploited the already available tags in the trees to automatically identify and label chunks in their Turkish translations. We used conditional random fields (CRF) to train a model over the annotated data. We report our results on different levels of chunk resolution.

In this paper, we analyze the errors of NP attachments that occur when they combine with NP's ending in possessor clitic.We suggest a simple pattern based detection and correction solution.We illustrate the errors with examples from Penn Treebank.

In this paper, we report our preliminary efforts in building an English-Turkish parallel treebank corpus for statistical machine translation. In the corpus, we manually generated parallel trees for about 5,000 sentences from Penn Treebank. English sentences in our set have a maximum of 15 tokens, including punctuation. We constrained the translated trees to the reordering of the children and the replacement of the leaf nodes with appropriate glosses. We also report the tools that we built and used in our tree translation task.

This paper presents the results of main part-of-speech tagging of Turkish sentences using Conditional Ran- dom Fields (CRFs). Although CRFs are applied to many different languages for part-of-speech (POS) tagging, Turkish poses interesting challenges to be modeled with them. The challenges include issues related to the statistical model of the problem as well as issues related to computational complexity and scaling. In this paper, we propose a novel model for main-POS tagging in Turkish. Furthermore, we pro- pose some approaches to reduce the computational complexity and allow better scaling characteristics or improve the performance without increased complexity. These approaches are discussed with respect to their advantages and disadvantages. We show that the best approach is competitive with the current state of the art in accuracy and also in training and test durations. The good results obtained imply a good first step towards full morphological disambiguation.