Ecological environment public opinion analysis method under river development

Abstract

The invention discloses an ecological environment public opinion analysis method under river development, and belongs to the field of river ecological environment public opinion analys.The method comprises the steps that an ecological environment related corpus is obtained, and associated words are obtained from the ecological environment related corpus based on river development project categories and serve as a retrieval keyword set of river development projects; retrieving the Internet information based on the retrieval keyword set, and obtaining the time-space information of the related speech and the speech publishing object; performing sentiment analysis on each related speech by using a sentiment analysis model to obtain a sentiment tendency of each related speech; and according to the time-space information of the speech publishing object and the emotional tendency of each related speech, performing fitting based on time to obtain the emotional tendency trend of each region about river development, and completing ecological environment public opinion analysis. The problems that an existing public opinion analysis method is insufficient in timeliness and does not consider regional characteristics are solved.

Description

Technical Field

[0001] The present invention belongs to the field of public opinion analysis of river ecological environment, and in particular relates to a public opinion analysis method of ecological environment under river development. Background Art

[0002] As environmental problems become increasingly prominent, the health of river ecological environment, as an important part of nature, is directly related to the sustainable development of human society. Therefore, the public's attention to river ecological environment is increasing, and related public opinion information is becoming increasingly rich and complex. In order to effectively respond to these public opinions, relevant departments and enterprises need to use advanced technical means to monitor and analyze.

[0003] Through public opinion monitoring and analysis technology, relevant departments can timely understand the public's attitude and views on the incident, as well as the development trend and scope of the incident. At the same time, combined with sentiment analysis technology, it can also discover the public's emotional tendencies towards certain remarks and behaviors in the incident, providing a reference for the handling and response of the incident.

[0004] The main difficulties faced by YX ladder development in cross-border ecological environment public opinion information distribution storage and feature knowledge base construction technology include: storing massive data requires huge storage space and efficient data management system, and building a feature knowledge base requires effective feature consistency association expression algorithm and robust semantic features. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a method for analyzing public opinion on the ecological environment under river development, which solves the problem that the existing public opinion analysis methods are not timely enough and do not take regional characteristics into consideration.

[0006] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: a method for analyzing public opinion on the ecological environment under river development, comprising the following steps:

[0007] S1. Obtain an ecological environment related corpus, and based on the category of river development projects, obtain associated words in the ecological environment related corpus as a search keyword set for river development projects;

[0008] S2, searching the Internet information based on the search keyword set, and obtaining the temporal and spatial information of the relevant speech and the object of the speech;

[0009] S3, using the sentiment analysis model to perform sentiment analysis on each relevant speech to obtain the sentiment tendency of each relevant speech;

[0010] S4. According to the spatiotemporal information of the subject of the speech and the sentiment tendency of each relevant speech, fitting is performed based on time to obtain the sentiment tendency trend of each region regarding river development and complete the ecological environment public opinion analysis.

[0011] Furthermore, the step S1 is specifically as follows:

[0012] S101, performing sentence processing on the corpus in the ecological environment related corpus, and removing the same sentences to obtain the sentence library related to the ecological environment;

[0013] S102, performing word segmentation processing on each sentence in the ecological environment related sentence library, and removing stop words to obtain the ecological environment related word library;

[0014] S103, using the Glove model, obtaining a word embedding vector of each word in the ecological environment related vocabulary;

[0015] S104, using the Glove model, obtaining a word embedding vector of the river development project category;

[0016] S105. Set a semantic relevance threshold, calculate the distance between the word embedding vector of each word in the ecological environment related vocabulary and the word embedding vector of the river development project category, and use the words with a distance greater than the semantic relevance threshold as associated words and add them to the search keyword set of river development projects.

[0017] Furthermore, the step S3 is specifically as follows:

[0018] S301, obtaining the current speech according to each relevant speech;

[0019] S302, dividing the current speech into sentences, obtaining the type of each sentence, and obtaining the current sentence;

[0020] S303, performing word segmentation processing on the current sentence, and obtaining the number of negative words, the number of positive sentiment words, and the number of negative sentiment words;

[0021] S304, obtaining the main word of the current sentence, and using the GloVe model to obtain the word embedding vector of the main word;

[0022] S305, obtaining the distance between the word embedding vector of the main word and the word embedding vector of each associated word in the search keyword set, and obtaining the associated word with the closest distance;

[0023] S306, obtaining the semantic tendency of the nearest associated word as the semantic tendency of the subject;

[0024] S307, using a sentiment analysis model to obtain the sentiment tendency of the current sentence according to the subject's semantic tendency, the number of negative words, the number of positive sentiment words, and the number of negative sentiment words;

[0025] S308, return to step S302, calculate the sentiment tendency of the next sentence, until all the sentences of the current speech are calculated, and then enter step S309;

[0026] S309, return to step S301 and start the calculation of the sentiment tendency of the next related speech, until the calculation of the sentiment tendency of the left and right related speeches is completed, and the sentiment tendency of each related speech is obtained.

[0027] Furthermore, the semantic inclination of the associated words in step S306 includes positive, negative and neutral.

[0028] Furthermore, the expression of the sentiment analysis model in step S307 is:

[0029]

[0030]

[0031] Among them, Q is the sentiment tendency of the current sentence; g is the semantic tendency of the subject; w i is the sentiment level of the ith positive sentiment word; S i is the ith positive sentiment word; n is the number of positive sentiment words; w r is the sentiment level of the rth negative sentiment word; T r is the rth negative sentiment word; q is the number of negative sentiment words; w j is the negation degree of the jth negative word; F j is the jth negative word; m is the number of negative sentiment words; γ is a binary function for judging emphasis and depression, which is 1 when the word is depressed first and then emphasized, otherwise, it is -1; ε is a neutral bias value, ranging from [0,1].

[0032] Furthermore, the step S4 is specifically as follows:

[0033] S401. According to the spatiotemporal information of the speech-making object, the relevant speeches made by the speech-making objects in each region are sorted out based on time to serve as spatiotemporal related river development project speech data:

[0034]

[0035] Among them, D z Develop engineering speech data for spatiotemporal river development in the zth region; t is time; is the sentiment tendency of speech in the zth region at time t; N z,t is the total number of relevant speeches in the zth region at time t; Q z,p,o is the sentiment tendency of the ath sentence of the pth related speech in the zth region; MAN() is the normalization function;

[0036] S402. Based on the spatial and temporal correlation of the speech data on river development projects, fitting is performed based on time to obtain the sentiment tendency trend of each region regarding river development and complete the ecological environment public opinion analysis.

[0037] The beneficial effects of the present invention are as follows: by collecting relevant comments through searching keywords, the real-time monitoring efficiency can be improved; by acquiring the sentiment tendency of sentences based on the differences in sentence patterns, the sentiment tendency analysis of relevant comments can be improved; at the same time, since public opinion analysis is performed based on region and time, the trend of public opinion changes can be better seen. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0039] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0040] like Figure 1 As shown, in one embodiment of the present invention, a method for analyzing public opinion on the ecological environment under river development includes the following steps:

[0041] S1. Obtain an ecological environment related corpus, and based on the category of river development projects, obtain associated words in the ecological environment related corpus as a search keyword set for river development projects;

[0042] S2, searching the Internet information based on the search keyword set, and obtaining the temporal and spatial information of the relevant speech and the object of the speech;

[0043] S3, using the sentiment analysis model to perform sentiment analysis on each relevant speech to obtain the sentiment tendency of each relevant speech;

[0044] S4. According to the spatiotemporal information of the subject of the speech and the sentiment tendency of each relevant speech, fitting is performed based on time to obtain the sentiment tendency trend of each region regarding river development and complete the ecological environment public opinion analysis.

[0045] The step S1 is specifically as follows:

[0046] S101, performing sentence processing on the corpus in the ecological environment related corpus, and removing the same sentences to obtain the sentence library related to the ecological environment;

[0047] S102, performing word segmentation processing on each sentence in the ecological environment related sentence library, and removing stop words to obtain the ecological environment related word library;

[0048] S103, using the Glove model, obtaining a word embedding vector of each word in the ecological environment related vocabulary;

[0049] S104, using the Glove model, obtaining a word embedding vector of the river development project category;

[0050] S105. Set a semantic relevance threshold, calculate the distance between the word embedding vector of each word in the ecological environment related vocabulary and the word embedding vector of the river development project category, and use the words with a distance greater than the semantic relevance threshold as associated words and add them to the search keyword set of river development projects.

[0051] In this embodiment, in order to monitor public opinion in real time, it is necessary to first select search keywords for screening relevant comments. In this embodiment, keywords with high similarity to river development projects are retrieved from the ecological environment related corpus as related words, and the relevance of words can be mapped to a high dimension to obtain word embedding vectors. The distance between vectors is the degree of relevance between words.

[0052] The step S3 is specifically as follows:

[0053] S301, obtaining the current speech according to each relevant speech;

[0054] S302, dividing the current speech into sentences, obtaining the type of each sentence, and obtaining the current sentence;

[0055] S303, performing word segmentation processing on the current sentence, and obtaining the number of negative words, the number of positive sentiment words, and the number of negative sentiment words;

[0056] S304, obtaining the main word of the current sentence, and using the GloVe model to obtain the word embedding vector of the main word;

[0057] S305, obtaining the distance between the word embedding vector of the main word and the word embedding vector of each associated word in the search keyword set, and obtaining the associated word with the closest distance;

[0058] S306, obtaining the semantic tendency of the nearest associated word as the semantic tendency of the subject;

[0059] S307, using a sentiment analysis model to obtain the sentiment tendency of the current sentence according to the subject's semantic tendency, the number of negative words, the number of positive sentiment words, and the number of negative sentiment words;

[0060] S308, return to step S302, calculate the sentiment tendency of the next sentence, until all the sentences of the current speech are calculated, and then enter step S309;

[0061] S309, return to step S301 and start the calculation of the sentiment tendency of the next related speech, until the calculation of the sentiment tendency of the left and right related speeches is completed, and the sentiment tendency of each related speech is obtained.

[0062] In this embodiment, the smallest unit of sentiment tendency is a word, but due to the particularity of sentence expression, different expressions of words with different sentiment tendencies will have different effects. Therefore, this embodiment uses sentences as the smallest unit to analyze sentiment tendency.

[0063] The semantic inclinations of the associated words in step S306 include positive, negative and neutral.

[0064] In this embodiment, the semantic tendency of the conjunction is the impact of the described event. For example, earthquake, flood, dam burst, etc. are negative tendencies, algae, river, hydrology, etc. are neutral tendencies, and irrigation, protection, environmental protection, etc. are positive tendencies.

[0065] The expression of the sentiment analysis model in step S307 is:

[0066]

[0067]

[0068] Among them, Q is the sentiment tendency of the current sentence; g is the semantic tendency of the subject; w i is the sentiment level of the ith positive sentiment word; S i is the ith positive sentiment word; n is the number of positive sentiment words; w r is the sentiment level of the rth negative sentiment word; T r is the rth negative sentiment word; q is the number of negative sentiment words; w j is the negation degree of the jth negative word; F j is the jth negative word; m is the number of negative sentiment words; γ is a binary function for judging emphasis and depression, which is 1 when the word is depressed first and then emphasized, otherwise, it is -1; ε is a neutral bias value, ranging from [0,1].

[0069] In this embodiment, calculation formulas for sentiment orientation are designed for commonly used sentence patterns, general declarative sentences, rhetorical questions and transitional sentences. The sentiment color of general declarative sentences is directly output, and in this case, the sentiment orientation is affected by the sentiment orientation of the words, and at the same time, the sentiment conversion function of negative words cannot be ignored.

[0070] The emotional tone of a rhetorical question is opposite to that of a general declarative sentence. Due to its sentence structure, its emotional expression has an additional layer of negation.

[0071] For transitional sentences, there are two ways of expressing emotions: first suppress and then praise, and first praise and then suppress. In this kind of expression, there is a part of negation in the praise, and there is also a part of negation in the suppression, and this part can be balanced through the association of negative words.

[0072] The step S4 is specifically as follows:

[0073] S401. According to the spatiotemporal information of the speech-making object, the relevant speeches made by the speech-making objects in each region are sorted out based on time to serve as spatiotemporal related river development project speech data:

[0074]

[0075] Among them, D z Develop engineering speech data for spatiotemporal river development in the zth region; t is time; is the sentiment tendency of speech in the zth region at time t; N z,t is the total number of relevant speeches in the zth region at time t; Q z,p,o is the sentiment tendency of the ath sentence of the pth related speech in the zth region; MAN() is the normalization function;

[0076] S402. Based on the spatial and temporal correlation of the speech data on river development projects, fitting is performed based on time to obtain the sentiment tendency trend of each region regarding river development and complete the ecological environment public opinion analysis.

[0077] In this embodiment, due to the different national conditions and people's cognition in different regions, different views on the development of the same river are often caused. Therefore, it is necessary to analyze based on regional public opinion. At the same time, based on time fitting, the trend of public opinion changes can be seen.

Claims

1. A method for analyzing public opinion on ecological environment under river development, characterized in that: The following steps are involved: S1. Obtain an ecological environment related corpus, and based on the category of river development projects, obtain associated words in the ecological environment related corpus as a search keyword set for river development projects; S2, searching the Internet information based on the search keyword set, and obtaining the time and space information of the relevant speech and the object of the speech; S3, using the sentiment analysis model to perform sentiment analysis on each relevant speech to obtain the sentiment tendency of each relevant speech; S4. According to the spatiotemporal information of the subject of the speech and the sentiment tendency of each relevant speech, fitting is performed based on time to obtain the sentiment tendency trend of each region regarding river development and complete the ecological environment public opinion analysis.

2. The method for analyzing public opinion on ecological environment under river development according to claim 1 is characterized in that: The step S1 is specifically as follows: S101, performing sentence processing on the corpus in the ecological environment related corpus, and removing the same sentences, to obtain an ecological environment related sentence library; S102, performing word segmentation processing on each sentence in the ecological environment related sentence library, and removing stop words to obtain the ecological environment related word library; S103, using the Glove model, obtaining a word embedding vector of each word in the ecological environment related vocabulary; S104, using the Glove model, obtaining a word embedding vector of the river development project category; S105. Set a semantic relevance threshold, calculate the distance between the word embedding vector of each word in the ecological environment related vocabulary and the word embedding vector of the river development project category, and use the words with a distance greater than the semantic relevance threshold as associated words and add them to the search keyword set of river development projects.

3. The method for analyzing public opinion on ecological environment under river development according to claim 1 is characterized in that: The step S3 is specifically as follows: S301, obtaining the current speech according to each related speech; S302, dividing the current speech into sentences, obtaining the type of each sentence, and obtaining the current sentence; S303, performing word segmentation processing on the current sentence, and obtaining the number of negative words, the number of positive sentiment words, and the number of negative sentiment words; S304, obtaining the main word of the current sentence, and using the GloVe model to obtain the word embedding vector of the main word; S305, obtaining the distance between the word embedding vector of the main word and the word embedding vector of each associated word in the search keyword set, and obtaining the associated word with the closest distance; S306, obtaining the semantic tendency of the nearest associated word as the semantic tendency of the subject; S307, using a sentiment analysis model to obtain the sentiment tendency of the current sentence according to the subject's semantic tendency, the number of negative words, the number of positive sentiment words, and the number of negative sentiment words; S308, return to step S302, calculate the sentiment tendency of the next sentence, until all the sentences of the current speech are calculated, and then enter step S309; S309, return to step S301 to start the calculation of the sentiment tendency of the next related speech, until the calculation of the sentiment tendency of the left and right related speeches is completed, and the sentiment tendency of each related speech is obtained.

4. The method for analyzing public opinion on ecological environment under river development according to claim 3 is characterized in that: The semantic inclinations of the associated words in step S306 include positive, negative and neutral.

5. The method for analyzing public opinion on ecological environment under river development according to claim 3 is characterized in that: The expression of the sentiment analysis model in step S307 is: Among them, Q is the sentiment tendency of the current sentence; g is the semantic tendency of the subject; w i is the sentiment level of the ith positive sentiment word; S i is the ith positive sentiment word; n is the number of positive sentiment words; w r is the sentiment level of the rth negative sentiment word; T r is the rth negative sentiment word; q is the number of negative sentiment words; w j is the negation degree of the jth negative word; F j is the jth negative word; m is the number of negative sentiment words; γ is a binary function for judging emphasis and depression, which is 1 when the word is depressed first and then emphasized, otherwise, it is -1; ε is a neutral bias value, ranging from [0,1].

6. The method for analyzing public opinion on ecological environment under river development according to claim 1 is characterized in that: The step S4 is specifically as follows: S401. According to the spatiotemporal information of the speech-making object, the relevant speeches made by the speech-making objects in each region are sorted out based on time to serve as spatiotemporal related river development project speech data: Among them, D z Develop engineering speech data for spatiotemporal river development in the zth region; t is time; is the sentiment tendency of speech in the zth region at time t; N z,t is the total number of relevant speeches in the zth region at time t; Q z,p,o is the sentiment tendency of the ath sentence of the pth related speech in the zth region; MAN() is the normalization function; S402. Based on the spatial and temporal correlation of river development project speech data, fitting is performed based on time to obtain the sentiment tendency trend of each region regarding river development and complete the ecological environment public opinion analysis.

Images