Article Preview
TopIntroduction
Improvements in information technologies alongside the development of Internet technology, the web has led to the formation of great size of information in the stack. The information contained in the web environment is usually present as text. Individually examination of the text in the web environment by humans is not possible in terms of time. Therefore, intensive studies have been conducted on the development of technology that can automatically evaluate these texts.
In order to evaluate the texts in the web environment, there is a requirement for intelligent systems. Understanding the texts of web environment by evaluating is not possible for the computers. Hence in attempt to evaluate the texts by the computers, new model developments that could estimate based upon the words contained in the text is a necessity. One of the important factors that determines the structure and the contents of the text is the words. It provides an opportunity in terms of evaluating structure and linage in sentence of words in the text. Estimating the types of words contained in the text with a model will ensure an important progress in the analysis of text content.
Different algorithms and models have been used in many studies to improve information extraction from text and text mining methods. He and Ling (2006) have used ontology-based metadata model by implementing the entity-relationship models and the classification of metadata information. Data mining methods are frequently used in the analysis of the text. Alhajj (2003) worked on an algorithm to perform mining on text documents by using the entity-relationship model. Direct access to regarding information from text documents, text filtering, text summary and data extraction can be achieved. In general, these operations expressed as text mining. Text mining methods are taken into consideration by many researchers. Tsai and Chang (2013) have used support vector machine algorithms for the model that takes into account the examples of selections. A large number of models can be administered to text classification. Ghiassi et al. (2012) worked on a dynamic neural network model for automatic text classification. Chen and Chen (2011) have benefited from the similarity measurements and chi-square statistics for text classification. Jiang et al. (2010) have worked on the text classification by using a feature extraction-based graph mining method. Klose et al. (2000) worked on document similarities on the text access subject.
Information extraction is one of the major areas of text mining. Different models have been developed on information extraction. Ko and Seo (2008) have obtained successful results in the operations such as text summarization and sentence extraction by benefitting from statistical techniques. Downey et al. (2010) worked on the uncontrolled data extraction from the web environment with the probabilistic model analysis. Data extraction models can be used on the texts in Turkish Language. Many researchers worked on algorithms and models that can extract information from texts in Turkish by using these methods (Tur et al., 2003; Tatar & Cicekli., 2011; Tatar, 2011; Adalı, 2009).
Stochastic model is one of the most commonly used methods in text mining. Stochastic models are widely used to provide information in various studies (Vlad et al., 2003; Hromic & Atkinson, 2012). Different statistical methods were used to investigate the morphological aspects of the words. Alajmi et al. (2011) studied morphological aspects of Arabic words by HMM. Papageorgiou (1994) offers different model from the current approaches for the segmentation of words in the Japanese Language by using HMM.
Since Turkish has an agglutinating language structure feature, words can take roots as well as suffixes. Meaning of the Turkish words can be changed by added suffixes. On the other hand, some suffixes do not cause any changes in the meaning of the word.