Semantic lexicons of English nouns for classification

Abstract

Sentiment classification is studied for a long time and there are many applications and many researches to service communities, commerce, politics, etc. In this research, we propose a new model to calculate the emotional values (or semantic scores) of English terms (English verbs, English nouns, English adjectives, English adverbs, etc.) as: first of all, we build our basis English sentiment dictionary (called bESD) by using Tanimoto Coefficient (Tanimoto measure, called TC) through Google search engine with AND operator and OR operator and then, we create many English noun phrases based on the English grammars (the English characteristics) and the valences of the English noun phrases are identified by their specific contexts. The English noun phrases often bring the semantics which the values (or emotional scores) are not fixed and are changed when they appear in their different contexts. Therefore, the results of the sentiment classification are not high accuracy if the English noun phrases bring the emotions and their semantic values (or their sentimental scores) are not changed in any context. For those reasons, we propose many rules based on English language grammars to calculate the sentimental values of the English noun phrases bearing emotion in their specific contexts. The results of this work are widely used in applications and researches of the English semantic classification.

Appendices

Appendix

See Tables 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14.

Table 6 Comparisons of our results with the surveys in Efron (2004), Yuen et al. (2004), Chen and Chiu (2009), Wang and Araki (2007), Taboada et al. (2006), Cimiano and Wenderoth (2007), Lu et al. (2010), Voll and Taboada (2007), Kundi et al. (2015), Mao et al. (2014), Turney and Littman (2003), Turney (2002), Rothfels and Tibshirani (2010), Nadaf et al. (2015), Feng et al. (2013), An and Hagiwara (2014) and Song et al. (2015)

Table 7 Comparisons of our model’s advantages and disadvantages with the surveys in Efron (2004), Yuen et al. (2004), Chen and Chiu (2009), Wang and Araki (2007), Taboada et al. (2006), Cimiano and Wenderoth (2007), Lu et al. (2010), Voll and Taboada (2007), Kundi et al. (2015), Mao et al. (2014), Turney and Littman (2003), Turney (2002), Rothfels and Tibshirani (2010), Nadaf et al. (2015), Feng et al. (2013), An and Hagiwara (2014) and Song et al. (2015)

Table 8 Types of pre-determiners

Table 9 Comparisons of our model’s results with the works related to the Tanimoto Coefficient (Tanimoto measure, called TC) in Kryszkiewicz (2014), Kryszkiewicz (2013), Kryszkiewicz and Podsiadly (2014), Awekar and Samatova (2009), Fligner et al. (2002), Kumar et al. (2010), Garavaglia (2001) and Kristensen and Pedersen (2010)

Table 10 Comparisons of our model’s advantage and disadvantages with the surveys related to the Tanimoto Coefficient (Tanimoto measure, called TC) in Kryszkiewicz (2014), Kryszkiewicz (2013), Kryszkiewicz and Podsiadly (2014), Awekar and Samatova (2009), Fligner et al. (2002), Kumar et al. (2010), Garavaglia (2001) and Kristensen and Pedersen (2010)

Table 11 Comparisons of our model’s results with the studies related to the sentiment classification, the semantic dictionary (or sentiment dictionary, or emotion lexicons, or semantic lexicons, or sentiment lexicons, etc.) as Manek et al. (2016), Agarwal and Mittal (2016a, b), Canuto et al. (2016), Ahmed and Danti (2016), Phu and Tuoi (2014), Tran et al. (2014), Godbole et al. (2007), Taboada et al. (2011), Nasukawa and Yi (2003), Choi and Cardie (2008), Qiu et al. (2009), Tan et al. (2008), Molinero et al. (2009), Andreevskaia and Bergler (2006), Steinberger et al. (2012), Brooke et al. (2009) and Remus et al. (2010)

Table 12 Comparisons of our model’s benefits and drawbacks with the surveys related to the sentiment classification, the semantic dictionary (or sentiment dictionary, or emotion lexicons, or semantic lexicons, or sentiment lexicons, etc.) as Manek et al. (2016), Agarwal and Mittal (2016a, b), Canuto et al. (2016), Ahmed and Danti (2016), Phu and Tuoi (2014), Tran et al. (2014), Godbole et al. (2007), Taboada et al. (2011), Nasukawa and Yi (2003), Choi and Cardie (2008), Qiu et al. (2009), Tan et al. (2008), Molinero et al. (2009), Andreevskaia and Bergler (2006), Steinberger et al. (2012), Brooke et al. (2009) and Remus et al. (2010)

Table 13 Comparisons of our model’s results with the researches related to the Tanimoto coefficient (TC) in emotional classification or in semantic classification in Bickerstaffe and Zukerman (2010), Deshpande et al. (2014), Gema Bello-Orgaz et al. (2014), Jadhao and Agrawal (2016) and Matthias Klusch and Patrick Kapahnke (2009)

Table 14 Comparisons of our model’s positives and negatives with the researches related to the Tanimoto coefficient (TC) in emotional classification or in semantic classification in Bickerstaffe and Zukerman (2010), Deshpande et al. (2014), Gema Bello-Orgaz et al. (2014), Jadhao and Agrawal (2016) and Matthias Klusch and Patrick Kapahnke (2009)

Appendices of all codes