Wong et al. [38], have used positive, negative, neutral classes, retweets and tweeters' information.
Several researchers such as Wang et al. [36] have considered positive, negative and neutral classes to extract the sentiment of a document (based on words and/or emoticons) and only few ones, such Khatua et al. [17], have examined the polarity degree (i.e. highly, moderately, weakly positive and negative classes).
The mean southerly aspects ranged from 226 229o (average aspect strength of 0.8), however the mean aspect of the neutral LMUs varied considerably because of the two neutral classes centered on 135o and 315o, with consequently a low average aspect strength of 0.05.
In fact, for the construction of the complete algorithm, we simply have to use Algorithms 1 to 4 (Unranking of neutral classes, atomic classes, disjoint unions and cartesian products, respectively) and Algorithm 6 (Unranking of weighted classes) given in [ 20] as subroutines.
Furthermore: Objects of size 0 are called neutral objects or tags and a class consisting of a single neutral object ϵ is called a neutral class, which will be denoted by ε (ε1, ε2,... to distinguish multiple neutral classes containing the objects ϵ1, ϵ2,..., respectively).
Tajima's D test (Table 3) did not reveal any significant departure from the neutral expectations and resulted in a slightly negative value for the dataset as a whole, and the muscat and neutral classes (D = -0.19, D = -0.35 and D = -0.42 respectively), and in a slightly positive value for the aromatic group (D = 0.67).
3-star reviews are used to prepare neutral class vectors.
({Accuracy},!pm,!1) ignores the Neutral class as it counts only severe errors (Leave vs. Remain).
The data were acquired during a blocked experiment design: viewing unpleasant (Class 1), neutral (Class 2) and pleasant pictures (Class 3).
Note also that SentiStrength and Sentiment140 present poor results in the 3-class experiments which can be explained by their bias to the neutral class as mentioned before.
The precision obtained by LNW for the neutral class was only 0.64, which indicates that the most common mistake made by the method, is misclassifying positive and negative messages as being neutral.
