Information Technology Reference
In-Depth Information
sentiment, and attitude. The combination of the work done for this sake is described
in the literature as opinion mining, sentiment analysis, and/or subjectivity analysis.
Determining the attitude of a speaker or a writer is of great importance for the
sentiment analysis as it is the main purpose of it. However, this can be very difficult at
times. In Li andWu's [ 17 ] own words: “The attitude can be any forms of judgment or
evaluation, the emotional state of the author when writing, or the intended emotional
communication.”
In sentiment analysis, two primary approaches are used, namely, linguistic and
machine learning. In linguistic approaches, studies are conducted by creating a set of
rules, and then by comparing themwith the analyzed text. An example of the linguistic
approach could be the study of Benamara et al. [ 4 ] who proposed a sentiment analysis
technique based on adverb-adjective combinations (AAC). The technique utilizes a
linguistic analysis of adverbs of degree.
Devitt and Ahmad [ 9 ] put forward that sentiment analysis in computational lin-
guistics has closely observed how textual features, such as lexical, syntactic, and
punctuation, alter the emotional content of the text. Furthermore, the sentiment analy-
sis considerably contributes to the automatic detection of these features so as to gather
a sentiment metric for a word, sentence, or the whole text.
On the other hand, in machine learning approaches, methods rely on statisti-
cal evaluations and analyzes such as frequency of positive and negative entities in
any text.
The reason behind the growing interest in this field stems from the benefits it can
provide. The main advantages of this research area are observed in stock market.
Predicting stock market behavior based on the sentiment results of Twitter posts,
according to Bollen et al. [ 7 ], can result in favorable outcomes. Moreover, O'Conner
et al. [ 21 ] underscore measuring public opinion poll in regards to presidential elec-
tions from blog data. Pang and Lee [ 22 ] mention the advantages of using the sen-
timent analysis in dealing with business intelligence tasks with respect to customer
feedback.
2.2.5 Text Classification
Text classification can be defined as assigning predefined category labels to docu-
ments such as e-mails to detect whether they are spam or nonspam, or web pages to
detect whether they are in English, German, or Turkish.
In this chapter, a supervised learning method was used, which is to say, first a
set of training documents were labeled, and then a machine learning algorithm was
applied to the document for classification.
Chen et al. [ 8 ] clearly state the increasing importance text classification. They
argue that the enhanced availability of digital texts and incremental increase in the
need to access them rendered text classification as a vital task. For a long time until
recently, various methods based on machine learning and statistical theory have been
implemented in text classification.
Search WWH ::




Custom Search