Public praise information evaluation method based on data analysis

Abstract

The invention discloses a public praise information evaluation method based on data analysis, and relates to the technical field of information evaluation. Content generated in a forum is acquired through a crawler technology; putting the original sentences into a filter, and obtaining official public praise aspects after screening of the filter; extracting adjective words, degree adverbs and negative words in an original sentence through an NLP technology, and evaluating the adjective words, the degree adverbs and the negative words; combining the emotion assignment of the degree adverb and the adjective to obtain a word-of-mouth evaluation positive degree score, and judging the word-of-mouth evaluation positive degree according to a scoring result; outputting the obtained public praise polarity and public praise evaluation positive degree; according to the method, the whole word-of-mouth mining task is divided into a plurality of sub-tasks to be split and simplified, the problems of data scarcity and marking quality are relieved to a great extent, and by calculating the word-of-mouth positive degree score, the positive degree of word-of-mouth evaluation can be judged while high accuracy is kept when more word-of-mouth information is extracted.

Description

Technical Field

[0001] The present invention relates to the technical field of information evaluation, and in particular to a word-of-mouth information evaluation method based on data analysis. Background Art

[0002] With the development of the Internet, online forums, as an important social media platform, have become an important channel for people to obtain and share information, exchange views and shape public opinion. Online forums contain a large amount of user-generated content, including evaluations and feedback on various products, services, events, and people. These contents reflect the real attitudes and preferences of users and have important reference value. Therefore, mining useful word-of-mouth information from online forums is of great significance in improving product quality, optimizing service levels, improving user experience, and monitoring public opinion trends.

[0003] However, word-of-mouth information in online forums has the following characteristics: 1. Massive: The amount of word-of-mouth information in online forums is huge, which is difficult to process and analyze manually; 2. Heterogeneous: The word-of-mouth information in online forums involves multiple fields and topics, and it is difficult to unify standards and norms; 3. Noise: The quality of word-of-mouth information in online forums is uneven, which is difficult to filter and correct; 4. Dynamic: The word-of-mouth information in online forums changes with time, events, users, etc., and is difficult to capture and update. Summary of the invention

[0004] The purpose of the present invention is to provide a word-of-mouth information evaluation method based on data analysis to solve the problems raised in the prior art.

[0005] To achieve the above object, the present invention provides the following technical solution: a word-of-mouth information evaluation method based on data analysis, the method comprising the following steps:

[0006] S1. Obtain the generated content in the forum through crawler technology, which is defined as the original sentence;

[0007] S2, putting the original sentence into the filter, and obtaining the official word-of-mouth aspects after being screened by the filter;

[0008] S3. Use NLP technology to extract adjective words from the original sentence, classify the adjectives and evaluate their emotional tendencies to obtain word-of-mouth opinions;

[0009] S4, extracting degree adverbs and negative words from the original sentence through NLP technology, and evaluating the degree of sentiment enhancement of the extracted degree adverbs;

[0010] S5. Score the positive degree of word-of-mouth evaluation by combining the emotional evaluation of degree adverbs and adjectives, and judge the positive degree of word-of-mouth evaluation according to the score results;

[0011] S6. Output the acquired word-of-mouth polarity and the degree of positivity of word-of-mouth evaluation.

[0012] Further, in step S1: crawling data from the target network forum using crawler technology to extract user-generated content; and storing the extracted user-generated content as original sentences in a local database;

[0013] Furthermore, in step S2: the original sentence is placed in a filter, and after filtering by parts of speech and stop words, it is recombined into nouns or noun groups, and the recombined nouns and noun phrases are defined as word-of-mouth candidate words, and the word-of-mouth candidate words are placed in a second filter, and the formal word-of-mouth terms are obtained after screening by the fine-tuned Bert model.

[0014] Further, in step S2: the following steps are included:

[0015] S2-1: Import the data required for part-of-speech tagging from NLTK Corpora by loading the python nltk library and import the stop word dataset;

[0016] S2-2: Perform POS Tagging after word segmentation on the original sentence to mark the part of speech of each word;

[0017] S2-3: After obtaining the part of speech of each word, select and filter out the parts of speech including 'CC', 'CD', 'DT', 'JJ', 'JJR', 'JJS', 'MD', 'RB', 'RBR', 'UH', 'VB', 'VBD', 'VBG', 'VBN', 'VBP', 'VBZ' and special symbols;

[0018] S2-4: Use the stop word dataset to perform another round of filtering on the results of the previous step;

[0019] S2-5: The remaining words are combined according to their positional relationship in the original sentence, and the combined words or phrases enter the candidate queue for word-of-mouth;

[0020] S2-6: A batch of data is manually labeled to fine-tune Bert, allowing it to calculate the probability that a certain word-of-mouth candidate word corresponds to the true word-of-mouth information in the original sentence. The input received by Bert is a combination of the word-of-mouth candidate word and the original sentence, and the output is the probability value p of being judged as word-of-mouth. If p is greater than z, it is considered that the input word-of-mouth candidate word corresponds to the word-of-mouth word in the original sentence.

[0021] Furthermore, in step S3: the original sentence is segmented by the Jieba library in the natural language processing technology, and the adjectives in the sentence are extracted and classified into three types: positive adjectives, negative adjectives and neutral adjectives. The sentiment tendency of the original sentence is determined according to the adjective classification, and the number of the three types of adjectives is counted and recorded respectively; a mark value r is taken to mark the positive and negative directions of the adjectives, and r is assigned in three cases, when the adjective is a positive adjective, r=1, when the adjective is a negative adjective, r=-1, when the adjective is a neutral adjective, r=0; the number of positive adjectives is used to represent positive word-of-mouth evaluation, and negative adjectives are used to represent negative word-of-mouth evaluation, and the number of adjectives and the sentiment assignment are used to calculate the word-of-mouth polarity.

[0022] Furthermore, in step S3: the Boson NLP sentiment dictionary is used to assign sentiment values ​​to adjectives of different categories, and whether the adjective can be used as a word-of-mouth opinion word is determined based on the sentiment value of the extracted adjective, and the sentiment value of each adjective is weighted averaged when the original sentence contains n adjectives, and the average sentiment value of the group of adjectives is calculated and extracted, and a sentiment value standard value s is set, and the sentiment value of the adjective or the average sentiment value of a group of adjectives is compared with the standard value s. When the sentiment value of the extracted adjective exceeds the standard value s, it is determined that the adjective can be used as a word-of-mouth opinion word.

[0023] Further, in step S4: the degree adverbs and negative words in the original sentence are extracted by natural language processing technology, and the degree adverbs are classified into levels, and the degree adverbs are graded using the scores q of the degree adverbs in the degree adverb dictionary, and the degree adverbs with scores less than 1 are defined as first-level degree adverbs, and the degree adverbs with scores greater than 1 are defined as second-level degree adverbs; and the degree to which the degree adverbs strengthen the emotional tendency is defined, and the first-level degree adverbs are defined as emotional weakening degree adverbs, and the second-level degree adverbs are defined as emotional strengthening degree adverbs.

[0024] Furthermore, in step S4: the number m of negative words in the original sentence is recorded, and after extracting the negative words from the original sentence, the negative words are combined with the sentiment tendency of the adjectives for analysis, and an odd number of negative words located before the adjectives can reverse the sentiment tendency of the adjectives.

[0025] Further, in step S5: the adjectives, degree adverbs and negative words extracted from the original sentence are combined to analyze the evaluation positivity of the original sentence, and a scoring method for the positivity of word-of-mouth evaluation is established. The positivity score of the word-of-mouth evaluation is calculated by superimposing the sentiment assignment of the adjectives and the scores of the degree adverbs of different levels, and the positivity score of the word-of-mouth evaluation is calculated using the formula:

[0026]

[0027] Among them, y represents the calculated positive degree score of word-of-mouth evaluation, q is the emotional strength score of the degree adverb, n is the number of adjectives, and x is the number of adjectives. i Represents the sentiment assignment of different adjectives, r is the adjective sentiment assignment mark value, r=1,0,-1, m represents the number of negative words in the original sentence; and it is stipulated that the positive degree score of word-of-mouth evaluation of positive adjectives is positive, the positive degree score of word-of-mouth evaluation of negative adjectives is negative, and the positive degree score of word-of-mouth evaluation of neutral adjectives is zero. When there are negative words, the negative words can reverse the positive degree score of word-of-mouth evaluation; through the established word-of-mouth positivity evaluation method, not only the polarity of word-of-mouth opinions can be obtained, but also whether the word-of-mouth evaluation is positive, neutral or negative can be judged, and the positivity of the word-of-mouth evaluation can be judged according to the scoring results.

[0028] Further, in step S6: according to the obtained word-of-mouth aspects and word-of-mouth opinions, as well as the original original sentence, three combinations of "word-of-mouth aspects + original sentences", "word-of-mouth aspects + word-of-mouth opinions", and "word-of-mouth aspects + original sentences + word-of-mouth opinions" are arranged, and Bert is fine-tuned according to each combination, the input is the current combination, the probabilities of three labels are output, and the combination with the highest accuracy among the three combinations is selected as the final solution, and the label calculated by Bert is the word-of-mouth polarity corresponding to the current word-of-mouth aspect; and the calculated word-of-mouth evaluation positivity score is output to obtain the evaluation result of the word-of-mouth information. There are three cases of the calculated evaluation positivity score, namely y>0, y=0 and y<0, which correspond to the positive, neutral and negative three cases of the word-of-mouth evaluation positivity respectively.

[0029] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention divides the entire word-of-mouth mining task into multiple subtasks for splitting and simplification. For the use of Bert, only classification tasks are performed in each subtask instead of generation tasks. In this way, only less training data is needed to achieve a higher accuracy rate, which greatly alleviates the problems of data scarcity and annotation quality. Moreover, through the word-of-mouth positivity scoring, it is possible to maintain a high accuracy rate while extracting more word-of-mouth information while judging the positivity of the word-of-mouth evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 The figure is a flow chart of the method for evaluating word-of-mouth information based on data analysis of the present invention. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] like Figure 1 As shown, the present invention provides a technical solution, a word-of-mouth information evaluation method based on data analysis, the method comprising the following steps:

[0033] S1. Obtain the generated content in the forum through crawler technology, which is defined as the original sentence;

[0034] S2, putting the original sentence into the filter, and obtaining the official word-of-mouth aspects after being screened by the filter;

[0035] S3. Use NLP technology to extract adjective words from the original sentence, classify the adjectives and evaluate their emotional tendencies to obtain word-of-mouth opinions;

[0036] S4, extracting degree adverbs and negative words from the original sentence through NLP technology, and evaluating the degree of sentiment enhancement of the extracted degree adverbs;

[0037] S5. Score the positive degree of word-of-mouth evaluation by combining the emotional evaluation of degree adverbs and adjectives, and judge the positive degree of word-of-mouth evaluation according to the score results;

[0038] S6. Output the acquired word-of-mouth polarity and the degree of positivity of word-of-mouth evaluation.

[0039] In step S1: crawling data from the target network forum using crawler technology to extract user-generated content; and storing the extracted user-generated content as original sentences in a local database;

[0040] In step S2: the original sentence is placed in the filter, and after filtering by part of speech and stop words, it is recombined into nouns or noun groups, and the recombined nouns and noun phrases are defined as word-of-mouth candidate words. The word-of-mouth candidate words are placed in the second filter, and the formal word-of-mouth words are obtained after screening by the fine-tuned Bert model.

[0041] In step S2: the following steps are included:

[0042] S2-1: Import the data required for part-of-speech tagging from NLTK Corpora by loading the python nltk library and import the stop word dataset;

[0043] S2-2: Perform POS Tagging after word segmentation on the original sentence to mark the part of speech of each word;

[0044] S2-3: After obtaining the part of speech of each word, select and filter out the parts of speech including 'CC', 'CD', 'DT', 'JJ', 'JJR', 'JJS', 'MD', 'RB', 'RBR', 'UH', 'VB', 'VBD', 'VBG', 'VBN', 'VBP', 'VBZ' and special symbols;

[0045] S2-4: Use the stop word dataset to perform another round of filtering on the results of the previous step;

[0046] S2-5: The remaining words are combined according to their positional relationship in the original sentence, and the combined words or phrases enter the candidate queue for word-of-mouth;

[0047] S2-6: A batch of data is manually labeled to fine-tune Bert, allowing it to calculate the probability that a certain word-of-mouth candidate word corresponds to the true word-of-mouth information in the original sentence. The input received by Bert is a combination of the word-of-mouth candidate word and the original sentence, and the output is the probability value p of being judged as word-of-mouth. If p is greater than z, it is considered that the input word-of-mouth candidate word corresponds to the word-of-mouth word in the original sentence.

[0048] In step S3: the original sentence is segmented by Jieba library in natural language processing technology, and the adjectives in the sentence are extracted and classified into three types: positive adjectives, negative adjectives and neutral adjectives. The sentiment tendency of the original sentence is determined according to the adjective classification, and the number of the three types of adjectives is counted and recorded respectively; a mark value r is taken to mark the positive and negative directions of the adjectives, and r is assigned in three cases, when the adjective is a positive adjective, r=1, when the adjective is a negative adjective, r=-1, when the adjective is a neutral adjective, r=0; the number of positive adjectives is used to represent positive word-of-mouth evaluation, and negative adjectives are used to represent negative word-of-mouth evaluation. The number of adjectives and the sentiment assignment are used to calculate the word-of-mouth polarity.

[0049] In step S3: use the Boson NLP sentiment dictionary to assign sentiment to adjectives of different categories, judge whether the adjective can be used as a word-of-mouth opinion word based on the sentiment assignment of the extracted adjective, and take a weighted average of the sentiment assignment of each adjective when the original sentence contains n adjectives, calculate and extract the average sentiment assignment of the group of adjectives, set the sentiment assignment standard value s, compare the sentiment assignment of the adjective or the average sentiment assignment of a group of adjectives with the standard value s, and when the sentiment assignment of the extracted adjective exceeds the standard value s, it is judged that the adjective can be used as a word-of-mouth opinion word.

[0050] In step S4: the degree adverbs and negative words in the original sentence are extracted by natural language processing technology, and the degree adverbs are classified into levels. The degree adverbs are graded using the scores q of the degree adverbs in the degree adverb dictionary, and the degree adverbs with scores less than 1 are defined as first-level degree adverbs, and the degree adverbs with scores greater than 1 are defined as second-level degree adverbs; and the degree to which the degree adverbs strengthen the emotional tendency is defined, and the first-level degree adverbs are defined as emotional weakening degree adverbs, and the second-level degree adverbs are defined as emotional strengthening degree adverbs.

[0051] In step S4: record the number m of negative words in the original sentence, and after extracting the negative words from the original sentence, combine the negative words with the sentiment tendency of the adjectives for analysis. An odd number of negative words located before the adjectives can reverse the sentiment tendency of the adjectives.

[0052] In step S5: the adjectives, degree adverbs and negative words extracted from the original sentence are combined to analyze the evaluation positivity of the original sentence, and a scoring method for the positivity of word-of-mouth evaluation is established. The positivity score of the word-of-mouth evaluation is calculated by superimposing the sentiment assignment of the adjectives and the scores of the degree adverbs of different levels, and the positivity score of the word-of-mouth evaluation is calculated using the formula:

[0053]

[0054] Among them, y represents the calculated positive degree score of word-of-mouth evaluation, q is the emotional strength score of the degree adverb, n is the number of adjectives, and x is the number of adjectives. i Represents the sentiment assignment of different adjectives, r is the adjective sentiment assignment mark value, r=1,0,-1, m represents the number of negative words in the original sentence; and it is stipulated that the positive degree score of word-of-mouth evaluation of positive adjectives is positive, the positive degree score of word-of-mouth evaluation of negative adjectives is negative, and the positive degree score of word-of-mouth evaluation of neutral adjectives is zero. When there are negative words, the negative words can reverse the positive degree score of word-of-mouth evaluation; through the established word-of-mouth positivity evaluation method, not only the polarity of word-of-mouth opinions can be obtained, but also whether the word-of-mouth evaluation is positive, neutral or negative can be judged, and the positivity of the word-of-mouth evaluation can be judged according to the scoring results.

[0055] In step S6: according to the obtained word-of-mouth aspects and word-of-mouth opinions, as well as the original sentence, three combinations of "word-of-mouth aspects + original sentences", "word-of-mouth aspects + word-of-mouth opinions", and "word-of-mouth aspects + original sentences + word-of-mouth opinions" are arranged, and Bert is fine-tuned according to each combination, the input is the current combination, the probabilities of three labels are output, and the combination with the highest accuracy among the three combinations is selected as the final solution, and the label calculated by Bert is the word-of-mouth polarity corresponding to the current word-of-mouth aspect; and the calculated word-of-mouth evaluation positivity score is output to obtain the evaluation result of the word-of-mouth information. There are three cases of the calculated evaluation positivity score, namely y>0, y=0 and y<0, which correspond to the positive, neutral and negative three cases of the word-of-mouth evaluation positivity respectively.

[0056] Embodiment 1:

[0057] In step S1: crawling data from the target network forum using crawler technology to extract user-generated content; and storing the extracted user-generated content as original sentences in a local database;

[0058] In step S2: the original sentence is placed in the filter, and after filtering by part of speech and stop words, it is recombined into nouns or noun groups, and the recombined nouns and noun phrases are defined as word-of-mouth candidate words. The word-of-mouth candidate words are placed in the second filter, and the formal word-of-mouth words are obtained after screening by the fine-tuned Bert model.

[0059] The following steps are involved:

[0060] S2-1: Import the data required for part-of-speech tagging from NLTK Corpora by loading the python nltk library and import the stop word dataset;

[0061] S2-2: Perform POS Tagging after word segmentation on the original sentence to mark the part of speech of each word;

[0062] S2-3: After obtaining the part of speech of each word, select and filter out parts of speech such as 'CC', 'CD', 'DT', 'JJ', 'JJR', 'JJS', 'MD', 'RB', 'RBR', 'UH', 'VB', 'VBD', 'VBG', 'VBN', 'VBP', 'VBZ' and a series of special symbols;

[0063] S2-4: Use the stop word dataset to perform another round of filtering on the results of the previous step;

[0064] S2-5: The remaining words are combined according to their positional relationship in the original sentence, and the combined words or phrases enter the candidate queue for word-of-mouth;

[0065] S2-6: A batch of data is manually labeled to fine-tune Bert, allowing it to calculate the probability that a certain word-of-mouth candidate word corresponds to the true word-of-mouth information in the original sentence. The input received by Bert is a combination of the word-of-mouth candidate word and the original sentence, and the output is the probability value p of being judged as word-of-mouth. If p is greater than z=0.6, it is considered that the input word-of-mouth candidate word corresponds to the word-of-mouth word in the original sentence.

[0066] In step S3: the original sentence is segmented by Jieba library in natural language processing technology, and the adjectives in the sentence are extracted and classified into three types: positive adjectives, negative adjectives and neutral adjectives. The sentiment tendency of the original sentence is determined according to the adjective classification, and the number of the three types of adjectives is counted and recorded respectively; a mark value r is taken to mark the positive and negative directions of the adjectives, and r is assigned in three cases, when the adjective is a positive adjective, r=1, when the adjective is a negative adjective, r=-1, when the adjective is a neutral adjective, r=0; the number of positive adjectives is used to represent positive word-of-mouth evaluation, and negative adjectives are used to represent negative word-of-mouth evaluation. The number of adjectives and the sentiment assignment are used to calculate the word-of-mouth polarity.

[0067] In step S3: use the Boson NLP sentiment dictionary to assign sentiment to adjectives of different categories, judge whether the adjective can be used as a word-of-mouth opinion word based on the sentiment assignment of the extracted adjective, and take a weighted average of the sentiment assignment of each adjective when the original sentence contains n adjectives, calculate and extract the average sentiment assignment of the group of adjectives, and set the sentiment assignment standard value s=0.9, compare the sentiment assignment of the adjective or the average sentiment assignment of a group of adjectives with the standard value s=0.9, and when the sentiment assignment of the extracted adjective exceeds the standard value s=0.9, it is judged that the adjective can be used as a word-of-mouth opinion word.

[0068] In step S4: the degree adverbs and negative words in the original sentence are extracted by natural language processing technology, and the degree adverbs are classified into levels. The degree adverbs are graded using the scores q of the degree adverbs in the degree adverb dictionary, and the degree adverbs with scores less than 1 are defined as first-level degree adverbs, and the degree adverbs with scores greater than 1 are defined as second-level degree adverbs; and the degree to which the degree adverbs strengthen the emotional tendency is defined, and the first-level degree adverbs are defined as emotional weakening degree adverbs, and the second-level degree adverbs are defined as emotional strengthening degree adverbs.

[0069] In step S4: record the number m of negative words in the original sentence, and after extracting the negative words from the original sentence, combine the negative words with the sentiment tendency of the adjectives for analysis. An odd number of negative words located before the adjectives can reverse the sentiment tendency of the adjectives.

[0070] In step S5: the adjectives, degree adverbs and negative words extracted from the original sentence are combined to analyze the evaluation positivity of the original sentence, and a scoring method for the positivity of word-of-mouth evaluation is established. The positivity score of the word-of-mouth evaluation is calculated by superimposing the sentiment assignment of the adjectives and the scores of the degree adverbs of different levels, and the positivity score of the word-of-mouth evaluation is calculated using the formula:

[0071]

[0072] Among them, y represents the calculated positive degree score of word-of-mouth evaluation, q is the emotional strength score of the degree adverb, n is the number of adjectives, and x is the number of adjectives. i Represents the sentiment assignment of different adjectives, r is the adjective sentiment assignment mark value, r = 1, 0, -1, m represents the number of negative words in the original sentence; and stipulates that the positive degree of word-of-mouth evaluation of positive adjectives is positive, the positive degree of word-of-mouth evaluation of negative adjectives is negative, and the positive degree of word-of-mouth evaluation of neutral adjectives is zero. When there are negative words, the negative words can reverse the positive degree of word-of-mouth evaluation; through the established word-of-mouth positive degree evaluation method, not only can the polarity of word-of-mouth opinions be obtained, but also the word-of-mouth evaluation can be judged to be positive, neutral or negative, and the positive degree of word-of-mouth evaluation can be judged according to the scoring results;

[0073] In step S6: according to the obtained word-of-mouth aspects and word-of-mouth opinions, as well as the original sentence, three combinations are arranged: "word-of-mouth aspects + original sentences", "word-of-mouth aspects + word-of-mouth opinions", and "word-of-mouth aspects + original sentences + word-of-mouth opinions". Bert is fine-tuned according to each combination, and the input is the current combination. The probabilities of the three labels are output. The combination with the highest accuracy among the three combinations is selected as the final solution, and the label calculated by Bert is the word-of-mouth polarity corresponding to the current word-of-mouth aspect. In our experiment, we found that the combination of "word-of-mouth aspects + original sentences" has the highest accuracy, which may be due to :Word-of-mouth opinions can reflect the author's emotional tendencies very concisely and quickly, but at the cost of discarding a large amount of contextual information, so word-of-mouth opinions have a certain one-sidedness, which may mislead the actual word-of-mouth polarity judgment; and after scoring the information of the original sentence through the evaluation positivity scoring method, the evaluation result of the word-of-mouth information is obtained and the evaluation result of the word-of-mouth information is output. There are three cases of the calculated evaluation positivity score, namely y>0, y=0 and y<0, which correspond to the positive, neutral and negative three situations of the word-of-mouth evaluation positivity, and the higher the positivity of the word-of-mouth evaluation, the higher the positivity of the word-of-mouth information.

[0074] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A word-of-mouth information evaluation method based on data analysis, characterized in that: The following steps are involved: S1. Obtain the generated content in the forum through crawler technology, which is defined as the original sentence; S2, putting the original sentence into the filter, and obtaining the official word-of-mouth aspects after being screened by the filter; S3. Use NLP technology to extract adjective words from the original sentence, classify the adjectives and evaluate their emotional tendencies to obtain word-of-mouth opinions; S4, extracting degree adverbs and negative words from the original sentence through NLP technology, and evaluating the degree of sentiment enhancement of the extracted degree adverbs; S5. Score the positive degree of word-of-mouth evaluation by combining the emotional evaluation of degree adverbs and adjectives, and judge the positive degree of word-of-mouth evaluation according to the score results; S6. Output the acquired word-of-mouth polarity and the degree of positivity of word-of-mouth evaluation.

2. The word-of-mouth information evaluation method based on data analysis according to claim 1 is characterized in that: In step S1: crawling data from the target network forum using crawler technology to extract user-generated content; and storing the extracted user-generated content as original sentences in a local database.

3. The word-of-mouth information evaluation method based on data analysis according to claim 1 is characterized in that: In step S2: the original sentence is placed in the filter, and after filtering by part of speech and stop words, it is recombined into nouns or noun groups, and the recombined nouns and noun phrases are defined as word-of-mouth candidate words. The word-of-mouth candidate words are placed in the second filter, and the formal word-of-mouth words are obtained after screening by the fine-tuned Bert model.

4. The word-of-mouth information evaluation method based on data analysis according to claim 1 is characterized in that: In step S2: the following steps are included: S2-1: Import the data required for part-of-speech tagging from NLTK Corpora by loading the python nltk library and import the stop word dataset; S2-2: Perform POS Tagging on the original sentence after word segmentation to mark the part of speech of each word; S2-3: After obtaining the part of speech of each word, filter out the parts of speech including 'CC', 'CD', 'DT', 'JJ', 'JJR', 'JJS', 'MD', 'RB', 'RBR', 'UH', 'VB', 'VBD', 'VBG', 'VBN', 'VBP', 'VBZ' and special symbols; S2-4: Use the stop word dataset to perform another round of filtering on the results of the previous step; S2-5: The remaining words are combined according to their positional relationship in the original sentence, and the combined words or phrases enter the candidate queue for word-of-mouth; S2-6: A batch of data is manually labeled to fine-tune Bert, allowing it to calculate the probability that a certain word-of-mouth candidate word corresponds to the true word-of-mouth information in the original sentence. The input received by Bert is a combination of the word-of-mouth candidate word and the original sentence, and the output is the probability value p of being judged as word-of-mouth. If p is greater than z, it is considered that the input word-of-mouth candidate word corresponds to the word-of-mouth word in the original sentence.

5. The word-of-mouth information evaluation method based on data analysis according to claim 1 is characterized in that: In step S3: the original sentence is segmented by Jieba library in natural language processing technology, and the adjectives in the sentence are extracted and classified into three types: positive adjectives, negative adjectives and neutral adjectives. The sentiment tendency of the original sentence is determined according to the adjective classification, and the number of the three types of adjectives is counted and recorded respectively; a mark value r is taken to mark the positive and negative directions of the adjectives, and r is assigned in three cases, when the adjective is a positive adjective, r=1, when the adjective is a negative adjective, r=-1, when the adjective is a neutral adjective, r=0.

6. The word-of-mouth information evaluation method based on data analysis according to claim 1 is characterized in that: In step S3: use the Boson NLP sentiment dictionary to assign sentiment to adjectives of different categories, judge whether the adjective can be used as a word-of-mouth opinion word based on the sentiment assignment of the extracted adjective, and take a weighted average of the sentiment assignment of each adjective when the original sentence contains n adjectives, calculate and extract the average sentiment assignment of the group of adjectives, set the sentiment assignment standard value s, compare the sentiment assignment of the adjective or the average sentiment assignment of a group of adjectives with the standard value s, and when the sentiment assignment of the extracted adjective exceeds the standard value s, it is judged that the adjective can be used as a word-of-mouth opinion word.

7. The word-of-mouth information evaluation method based on data analysis according to claim 1 is characterized in that: In step S4: the degree adverbs and negative words in the original sentence are extracted by natural language processing technology, and the degree adverbs are classified into levels. The degree adverbs are graded using the scores q of the degree adverbs in the degree adverb dictionary, and the degree adverbs with scores less than 1 are defined as first-level degree adverbs, and the degree adverbs with scores greater than 1 are defined as second-level degree adverbs; and the degree to which the degree adverbs strengthen the emotional tendency is defined, and the first-level degree adverbs are defined as emotional weakening degree adverbs, and the second-level degree adverbs are defined as emotional strengthening degree adverbs.

8. The word-of-mouth information evaluation method based on data analysis according to claim 1 is characterized by: In step S4: record the number m of negative words in the original sentence, and after extracting the negative words from the original sentence, combine the negative words with the sentiment tendency of the adjectives for analysis. An odd number of negative words located before the adjectives can reverse the sentiment tendency of the adjectives.

9. The word-of-mouth information evaluation method based on data analysis according to claim 1, characterized in that: In step S5: the adjectives, degree adverbs and negative words extracted from the original sentence are combined to analyze the evaluation positivity of the original sentence, and a scoring method for the positivity of word-of-mouth evaluation is established. The positivity score of the word-of-mouth evaluation is calculated by superimposing the sentiment assignment of the adjectives and the scores of the degree adverbs of different levels, and the positivity score of the word-of-mouth evaluation is calculated using the formula: Among them, y represents the calculated positive degree score of word-of-mouth evaluation, q is the emotional strength score of the degree adverb, n is the number of adjectives, and x is the number of adjectives. i Represents the sentiment assignment of different adjectives, r is the adjective sentiment assignment tag value, r = 1, 0, -1, and m represents the number of negative words in the original sentence.

10. The word-of-mouth information evaluation method based on data analysis according to claim 1 is characterized in that: In step S6: according to the obtained word-of-mouth aspects and word-of-mouth opinions, as well as the original original sentence, three combinations of "word-of-mouth aspects + original sentences", "word-of-mouth aspects + word-of-mouth opinions", and "word-of-mouth aspects + original sentences + word-of-mouth opinions" are arranged, and Bert is fine-tuned according to each combination, the input is the current combination, the probabilities of three labels are output, and the combination with the highest accuracy among the three combinations is selected as the final solution, and the label calculated by Bert is the word-of-mouth polarity corresponding to the current word-of-mouth aspect; and the calculated word-of-mouth evaluation positivity score is output to obtain the evaluation result of the word-of-mouth information. There are three cases of the calculated evaluation positivity score, namely y>0, y=0 and y<0, which correspond to the positive, neutral and negative three cases of the word-of-mouth evaluation positivity respectively.

Images