Semantic relations are an important part of the emotional
semantic dictionary. The semantic dictionary structurally stores
the relationship between vocabulary, and generally uses
different markers to store different semantic relationships
between vocabulary, such as lower-level relationship and
synonymous relationship.The existing influential sentiment
dictionaries are mainly WordNet and Hownet. These two
dictionaries are often used as the basic dictionary for
constructing other sentiment dictionaries.WordNet is a very
influential dictionary. It is a cognitive linguistics-based English
dictionary designed by psychologists, linguists and software
engineers at Princeton University [6].According to the meaning
of the word, nouns, verbs, adjectives and adverbs are each
formed into a synonym network, and there is a relationship
between the networks [3].HowNet is a description of the
concepts of words in Chinese and English, expressing the
relationship between concepts and concepts and the
relationships between the attributes of concepts [7]. WordNet
only considers the semantic, antisense, superior and
subordinate relationships between semantic class attributes. In
addition to considering the relationship of WordNet annotation,
HowNet also examined the relationship between attributes and
components [8]. Due to the large number of investigations and
the difficulty of construction, the vocabulary is very limited [3].
Turney et al. [9] proposed an emotional divergence
discrimination algorithm based on point mutual information
(PMI). The algorithm effect depends on the size of the
emotional word set, and the calculation amount is too large, so
the efficiency is not high. Kamps et al. [10]use the synonymous
relationship in WordNet to calculate the distance of the word to
be recognized to “good” and “bad” to determine its emotional
polarity. Words that are not associated with these two words
cannot be distinguished, and WordNet vocabulary is limited .
Zhu Yanlan et al.[11] calculated semantic bias based on
HowNet's semantic similarity and semantic correlation field
functions. Liu Weiping [12] proposed a method for calculating
the emotional weight of Chinese emotional words. Based on
the HowNet sentiment dictionary, the Chinese basic emotion
dictionary is constructed, and the Chinese text tendency is
judged according to this. Zhou Yongmei et al.Rao and
Ravichandran [13] used the connection between the synonym
and the superscript of WordNet to construct the edges between
the words, thus forming a picture. And use the graph-based
semi-supervised learning algorithm to classify words as
derogatory or derogatory.[14]built a Chinese sentiment
dictionary based on HowNet and SentiWordNet. These articles
are based on existing sentiment dictionaries to construct
Chinese sentiment dictionaries and perform emotional
calculations, or directly use existing dictionaries for emotional
calculations. However, it cannot solve the impact of
unregistered words on emotional calculations.
It can be seen from the above research that the existing
sentiment dictionary are mainly edited manually or semi-
automatically, resulting in the inability to update new semantic
relationships in a timely and rapid manner.Emotional
dictionary lacks the extraction of semantic similarity relations
and semantic strength relations. This article uses the
authoritative Modern Chinese Dictionary and Baidu
The extraction of the meaning relationship of emotional
words is based on the Modern Chinese Dictionary and the
Baidu interpretation of online crawling to form the basic
corpus. Their content is an explanation of the vocabulary and
the structure is similar.
The Modern Chinese Dictionary was written by the
Institute of Linguistics of the Chinese Academy of Social
Sciences [15]. This dictionary is characterized by normative,
scientific and practical. The book covers about 65,000 words,
which basically covers the current picture of the stable use of
Chinese vocabulary. The words, interpretations and examples
are all determined by a number of experts through investigation
and research. The inclusion and interpretation of the words in
the dictionary are constantly revised, reflecting the
development and changes of Chinese vocabulary in a timely
manner, and guaranteeing the high quality and authority of this
dictionary. Therefore, the vocabulary and its definitions in the
modern Chinese dictionary have a strong authority. The
semantic relationship of extracting emotional words based on
this dictionary also has certain authority [2]. In the Modern
Chinese Dictionary, the explained words, the interpretation of
the words, and the example sentences of the words constitute a
knowledge, as shown in Fig. 1.
In order to make the acquired emotional words more
comprehensive, the knowledge corpus has added Baidu
interpretation. Baidu Interpretation is a word explanation
captured from Baidu Encyclopedia.Baidu Encyclopedia is an
open, free network encyclopedia platform launched by
Baidu[16]. As of February 2018, Baidu Encyclopedia has
included more than 15.2 million entries, and more than 6.44
million netizens participated in the entry, covering almost all
known areas of knowledge. The content of the main body of
the entry introduces the entry according to a certain structure.
At the same time, in order to ensure the accuracy of the content,
the ideal state is to require reference materials for each piece of
content to supplement the evidence. Therefore, from the Baidu
Encyclopedia to grasp the meaning of words, its accuracy and
comprehensiveness also have certain guarantees.Each piece of
knowledge in Baidu interpretation consists of the interpretation
of words and words, as shown in Fig. 2. It is similar to the