“…Methods commonly used for sentiment words extraction include rule-based extraction (Bhamare and Prabhu, 2021; Zhang et al , 2019), part-of-speech-based extraction (Mukhtar and Khan, 2019; Yin et al , 2020) and N-gram extraction (Keshavarz and Abadeh, 2017; Salas-Zárate et al , 2017). Focusing on sentiment words selection and polarity judgement, a variety of solutions have been proposed, such as the linguistic rules-based method; the words similarity-based method, including point mutual information (Wang et al , 2020; Wei et al , 2017), context similarity (Yu et al , 2013) and semantic similarity information derived from Word2vec (Araque et al , 2018; Liu et al , 2019); the graph-based method (Sheng et al , 2014); and the classification-based method (Han et al , 2018). The effect of constructing a sentiment lexicon based on corpus is affected by many factors, such as the scale and quality of the corpus, the influence of candidate sentiment word extraction strategy, sentiment word selection and the sentiment tendency judgement strategy.…”