CN104572877A - Detection method and detection system of game public opinion - Google Patents
Detection method and detection system of game public opinion Download PDFInfo
- Publication number
- CN104572877A CN104572877A CN201410805964.3A CN201410805964A CN104572877A CN 104572877 A CN104572877 A CN 104572877A CN 201410805964 A CN201410805964 A CN 201410805964A CN 104572877 A CN104572877 A CN 104572877A
- Authority
- CN
- China
- Prior art keywords
- game
- word
- keyword
- network text
- classification
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/374—Thesaurus
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a detection method of game public opinion. The detection method comprises the following steps: acquiring public opinion data in a web text; segmenting words for the public opinion, and acquiring words in the web text; inquiring a preliminarily-established game key word dictionary, and acquiring a game key word in the words; calculating an emotional value of the game key word according to a preliminarily-established emotion dictionary; calculating the type of the web text by adopting a Bayes classification algorithm according to the game key word in the web text; counting the emotional value of the game key word and the quantity and type of the web text, and acquiring evaluation information of the game. Correspondently, the invention also discloses a detection system of the game public opinion. By adopting the method and system, the evaluation information of the game can be accurately acquired.
Description
Technical field
The present invention relates to Internet technical field, particularly relate to a kind of detection method and system of public sentiment of playing.
Background technology
In the exploitation and operation process of game, collecting the feedback information of game user to game by various mode is a very important job.By analyzing these information collected, not only can excavate the demand of player, understanding the hobby of game user, also can observe game user in the recent period to the attitude of certain game, to the hobby of game play.By means of these information, strategy formulation person can adjust the direction of game, and Game development teams can improve game content.Can say, the feedback information of game user is very important reference for a game team.UGC (User Generated Content) refers to that the user on internet produces content, in today of internet high speed development, contain a large amount of feedback informations in UGC, thus to the collection of public sentiment of playing in UGC, analyze and become the channel of most of game making team.
But today of internet data amount explosive growth, internet, while bringing mass data, also greatly adds the difficulty from these extracting data useful informations.The places such as prior art is the place of concentrating appearance at some game users, the game special edition of the forum of official of such as playing, mhkc or door, the content that manual read player delivers, sums up, analyzes and make report and be supplied to related personnel's reading.
But manual read needs the artificial of at substantial and time, and continuous surveillance needs constantly to read, analyze, and generates report, inefficiency.Meanwhile, because manpower is limited after all, can only read mass data sampling, make the correctness analyzed, the authority of report more depends on the level of people.And because UGC mostly is unstructured data, manual read is difficult to collect data and add up, and makes to lack statistical information useful in a large number in UGC in final result.
Summary of the invention
The embodiment of the present invention proposes a kind of detection method of public sentiment of playing and system, can the evaluation and test information of Obtaining Accurate game.
The embodiment of the present invention provides a kind of detection method of public sentiment of playing, and comprising:
Public sentiment data in collection network text;
Participle is carried out to described public sentiment data, obtains the word in described network text;
Inquire about the game keyword dictionary set up in advance, obtain the game keyword in described word;
According to the sentiment dictionary set up in advance, calculate the emotion value of described game keyword;
According to the game keyword in described network text, adopt Bayesian Classification Arithmetic, calculate the classification belonging to described network text;
Add up the emotion value of described game keyword, the quantity of described network text and classification, obtain the evaluation and test information of game.
Further, described participle is carried out to described public sentiment data, obtains the word in described network text, specifically comprise:
Based on game word dictionary and stop words dictionary, adopt stammerer participle, participle is carried out to described public sentiment data, obtains the word in described network text.
Further, the game keyword in described game keyword dictionary is comprised in described game word dictionary.
Further, the sentiment dictionary that described basis is set up in advance, calculates the emotion value of described game keyword, specifically comprises:
Inquire about the sentiment dictionary set up in advance, judge whether the modification word of described game keyword is emotion word; Described sentiment dictionary is the database of the weights storing multiple emotion word and correspondence thereof, described modification word is the word between the beginning of the sentence of the network text at described game keyword and its place, or described modification word is a upper word of playing between keyword in the network text at described game keyword and its place;
If described modification word is emotion word, then read the weights that described emotion word is corresponding;
According to described weights, calculate the emotion value of described game keyword.
Further, at the sentiment dictionary that described basis is set up in advance, after calculating the emotion value of described game keyword, also comprise:
All game keywords are classified, obtains game keyword categories;
According to the emotion value of game keyword each in described game keyword categories, calculate the comprehensive emotion value of described game keyword categories.
Further, described according to the game keyword in described network text, adopt Bayesian Classification Arithmetic to classify to described network text, obtain the classification of described network text, specifically comprise:
Based on the keyword classification training set generated in advance, the game keyword in described network text is classified, obtain the training sample classification of described network text;
Adopt Bayesian Classification Arithmetic, calculate the probability that described network text belongs to each training sample classification respectively, the training sample classification of maximum probability is the classification of described network text.
Preferably, the computing formula of described Bayesian Classification Arithmetic is as follows:
P(Ci|d)＝P(d|Ci)P(Ci)L(d,Ci)
Wherein, P (Ci │ d) belongs to the probability of training sample classification Ci for network text d, for there is the probability of described network text d in described training sample classification Ci in P (d │ Ci), P (Ci) is the marginal probability of described training sample classification Ci, L (d, Ci) be the length factor of described network text d, Len (d) is the number of word in described network text d, avgLen (Ci) is the word mean number of network text in described training sample classification Ci, and k is the disturbance degree of word number to described training sample classification Ci.
Accordingly, the embodiment of the present invention also provides a kind of detection system of public sentiment of playing, and comprising:
Acquisition module, for the public sentiment data in collection network text;
Word-dividing mode, for carrying out participle to described public sentiment data, obtains the word in described network text;
Keyword acquisition module, for inquiring about the game keyword dictionary set up in advance, obtains the game keyword in described word;
Emotion value computing module, for according to the sentiment dictionary set up in advance, calculates the emotion value of described game keyword;
Web text classification module, for according to the game keyword in described network text, adopts Bayesian Classification Arithmetic, calculates the classification belonging to described network text; And,
Evaluation and test data obtaining module, for adding up the emotion value of described game keyword, the quantity of described network text and classification, obtains the evaluation and test information of game.
Further, described word-dividing mode, specifically for based on game word dictionary and stop words dictionary, adopts stammerer participle, carries out participle, obtain the word in described network text to described public sentiment data.
Further, the game keyword in described game keyword dictionary is comprised in described game word dictionary.
Further, described emotion value computing module specifically comprises:
Judging unit, for inquiring about the sentiment dictionary set up in advance, judges whether the modification word of described game keyword is emotion word; Described sentiment dictionary is the database of the weights storing multiple emotion word and correspondence thereof, described modification word is the word between the beginning of the sentence of the network text at described game keyword and its place, or described modification word is a upper word of playing between keyword in the network text at described game keyword and its place;
Reading unit, during for judging described modification word as emotion word at described judging unit, reads the weights that described emotion word is corresponding; And,
Computing unit, for according to described weights, calculates the emotion value of described game keyword.
Further, the detection system of described game public sentiment also comprises:
Keyword classification module, for classifying to all game keywords, obtains game keyword categories; And,
Comprehensive emotion value computing module, for the emotion value according to game keyword each in described game keyword categories, calculates the comprehensive emotion value of described game keyword categories.
Further, described Web text classification module specifically comprises:
Keyword classification unit, for based on the keyword classification training set generated in advance, classifies to the game keyword in described network text, obtains the training sample classification of described network text; And,
Web text classification unit, for adopting Bayesian Classification Arithmetic, calculate the probability that described network text belongs to each training sample classification respectively, the training sample classification of maximum probability is the classification of described network text.
Preferably, the computing formula of described Bayesian Classification Arithmetic is as follows:
P(Ci|d)＝P(d|Ci)P(Ci)L(d,Ci)
Wherein, P (Ci │ d) belongs to the probability of training sample classification Ci for network text d, for there is the probability of described network text d in described training sample classification Ci in P (d │ Ci), P (Ci) is the marginal probability of described training sample classification Ci, L (d, Ci) be the length factor of described network text d, Len (d) is the number of word in described network text d, avgLen (Ci) is the word mean number of network text in described training sample classification Ci, and k is the disturbance degree of word number to described training sample classification Ci.
Implement the embodiment of the present invention, there is following beneficial effect:
The detection method of the game public sentiment that the embodiment of the present invention provides and system, can by the emotion value of keyword of playing in statistics public sentiment data, and the quantity of network text and classification, understands the Sentiment orientation of user to game, the evaluation and test information of Obtaining Accurate game; Based on game word dictionary and stop words dictionary, participle is carried out to public sentiment data, improve the accuracy rate to game proper noun participle, and avoid the redundancy keyword generation that has nothing to do in a large number; According to game keyword dictionary, obtain the game keyword in network text, the network text irrelevant with game content can be filtered out simultaneously, improve statistical efficiency and accuracy rate; Add up the emotion value of each game keyword, obtain the Sentiment orientation of user to game various piece; Word number in text Network Based, improves Bayesian Classification Arithmetic, and adopts the Bayesian Classification Arithmetic improved to classify to network text, improves accuracy rate and the recall rate of Web text classification.
Accompanying drawing explanation
Fig. 1 is the schematic flow sheet of an embodiment of the detection method of game public sentiment provided by the invention;
Fig. 2 is the structural representation of an embodiment of the detection system of game public sentiment provided by the invention.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, be clearly and completely described the technical scheme in the embodiment of the present invention, obviously, described embodiment is only the present invention's part embodiment, instead of whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art, not making the every other embodiment obtained under creative work prerequisite, belong to the scope of protection of the invention.
See Fig. 1, be the schematic flow sheet of an embodiment of the detection method of game public sentiment provided by the invention, comprise step S1 to step S6, specific as follows:
Public sentiment data in S1, collection network text;
S2, participle is carried out to described public sentiment data, obtain the word in described network text;
The game keyword dictionary that S3, inquiry are set up in advance, obtains the game keyword in described word;
The sentiment dictionary that S4, basis are set up in advance, calculates the emotion value of described game keyword;
S5, according to the game keyword in described network text, adopt Bayesian Classification Arithmetic, calculate the classification belonging to described network text;
S6, add up the emotion value of described game keyword, the quantity of described network text and classification, obtain the evaluation and test information of game.
Wherein, adopt the public sentiment data in reptile collection network text, reptile uses Scapy framework.When reptile gathers, first from root network address, crawl data, then the page of network text is entered, by URL (the Uniform Resource Locator of each network text, URL(uniform resource locator)) add URL queue, the connection in URL queue is resolved simultaneously, obtain the information of each network text, as network text title, content, author, deliver the time etc., and described information to be stored in a database.Wherein, network text comprises the model, the social webpage of good friend etc. in microblogging, forum.
Concrete, in step s 2, described participle is carried out to described public sentiment data, obtain the word in described network text, specifically comprise:
Based on game word dictionary and stop words dictionary, adopt stammerer participle, participle is carried out to described public sentiment data, obtains the word in described network text.
When carrying out participle to the public sentiment data collected, adopting stammerer participle, finding out the maximum cutting combination based on word frequency, improving participle accuracy rate.And optimize the dictionary of participle according to the actual requirements, add in dictionary by the proper noun in game, set up game word dictionary, it is more accurate to make the participle of game public sentiment data.Meanwhile, set up the stop words dictionary for network text data feature, based on stop words dictionary, the word after participle is screened, the generation of redundancy word of avoiding haveing nothing to do in a large number.
After completing participle, inquire about the game keyword dictionary set up in advance, from the word that participle obtains, find out game keyword, retain the network text with game keyword, filter out the network text irrelevant with game content, improve the efficiency that game public sentiment detects.
Further, the game keyword in described game keyword dictionary is comprised in described game word dictionary.
Concrete, in step s 4 which, the sentiment dictionary that described basis is set up in advance, calculates the emotion value of described game keyword, specifically comprises:
Inquire about the sentiment dictionary set up in advance, judge whether the modification word of described game keyword is emotion word; Described sentiment dictionary is the database of the weights storing multiple emotion word and correspondence thereof, described modification word is the word between the beginning of the sentence of the network text at described game keyword and its place, or described modification word is a upper word of playing between keyword in the network text at described game keyword and its place;
If described modification word is emotion word, then read the weights that described emotion word is corresponding;
According to described weights, calculate the emotion value of described game keyword.
Wherein, sentiment dictionary comprises basic emotion dictionary and colloquial style word dictionary, and use colloquial style word dictionary can improve the identification to colloquial style emotion word.The polarity of emotion word is divided into front, negative and neutral, and polarity is different, and the weights of emotion word are not identical yet.Meanwhile, in network text, emotion word has degree word and/or negative word is modified, and makes the weighted shared by each emotion word.Therefore, when judging the modification word of game keyword, inquiry emotion degree adverb dictionary and negative word dictionary while inquiry sentiment dictionary, make the calculating of the emotion value of game keyword more accurate.
The calculating of game keyword emotion value, need search forward it and modify word, and judge whether described modification word is emotion word, if described modification word is emotion word, then continue to search the degree word before described emotion word or negative word from game keyword.If have degree word before described emotion word, then the weights of emotion word are multiplied by the weights of degree word; If there is negative word before described emotion word, then the weights of emotion word are multiplied by-1.When finding the beginning of the sentence of network text at described game keyword place, or during a upper game keyword in the network text at described game keyword place, stop searching about described game keyword, obtain the emotion value of described game keyword.According to these computing method, calculate the emotion value obtaining each game keyword.
Further, at the sentiment dictionary that described basis is set up in advance, after calculating the emotion value of described game keyword, also comprise:
All game keywords are classified, obtains game keyword categories;
According to the emotion value of game keyword each in described game keyword categories, calculate the comprehensive emotion value of described game keyword categories.
After the emotion value obtaining each game keyword, according to the difference in functionality module in game, as equipment module, pet module, technical ability module etc., game keyword can be classified, and then obtains the comprehensive emotion value of this module.Based on the emotion value of each game keyword and the comprehensive emotion value of each game module, the Sentiment orientation of game user to game various piece can be obtained, be conducive to game plan personnel and carry out decision-making.
Concrete, in described step S5, described according to the game keyword in described network text, adopt Bayesian Classification Arithmetic to classify to described network text, obtain the classification of described network text, specifically comprise:
Based on the keyword classification training set generated in advance, the game keyword in described network text is classified, obtain the training sample classification of described network text;
Adopt Bayesian Classification Arithmetic, calculate the probability that described network text belongs to each training sample classification respectively, the training sample classification of maximum probability is the classification of described network text.
What the method for classifying to network text adopted is first classify to game keyword, then integrates according to the classification of game keyword and classify to network text.Wherein, game keyword is classified based on keyword classification training set, and keyword classification training set can pre-set according to demand, as the classification about game leak, about the classification etc. of game article.
In addition, owing to being mostly short text to the evaluation of game in network, in network text, the number of word is larger on the impact of network text classification, therefore, Bayesian Classification Arithmetic is improved, on the basis of traditional Bayesian Classification Arithmetic, consider the factor of word number in network text, realize classification to network text, thus the uncertainty controlling network text length is to the harmful effect of Bayesian Classification Arithmetic result, improves accuracy rate and the recall rate of classification.
Concrete, the computing formula of described Bayesian Classification Arithmetic is as follows:
P(Ci|d)＝P(d|Ci)P(Ci)L(d,Ci)
Wherein, P (Ci │ d) belongs to the probability of training sample classification Ci for network text d, for there is the probability of described network text d in described training sample classification Ci in P (d │ Ci), P (Ci) is the marginal probability of described training sample classification Ci, L (d, Ci) be the length factor of described network text d, Len (d) is the number of word in described network text d, avgLen (Ci) is the word mean number of network text in described training sample classification Ci, and k is the disturbance degree of word number to described training sample classification Ci.
Wherein, k is larger, and in network text d, the result of calculation impact of word number on P (Ci │ d) is larger.
Concrete, there is the probability of described network text d in described training sample classification Ci
the marginal probability of described training sample classification Ci
wherein, F (tj, Ci) is for having the network text quantity of eigenwert tj in described training sample classification Ci, F (Ci) is the quantity of network text in described training sample classification Ci, m is the quantity of training sample classification, and N is the total quantity of network text.
It should be noted that, on the basis of traditional Bayesian Classification Arithmetic, increase the length factor L (d of network text, Ci), realize the classification to network text, thus the uncertainty controlling network text length is to the harmful effect of Bayesian Classification Arithmetic result, improve accuracy rate and the recall rate of classification.
Further, game public sentiment data is added up, obtains the evaluation and test information of game.Statistics gatherer to the quantity of network text and network text in send the number of public sentiment data, the entirety that can obtain game enlivens situation, for development group holistic approach game pouplarity and traffic-operating period; The emotion value of statistics game keyword, obtains game user to the Sentiment orientation of each game keyword, is convenient to development of games group and finds potential point to be modified, improves game quality; According to functional module, game keyword is classified, and then adds up the emotion value of each functional module, obtain game user to the Sentiment orientation of each functional module, analyze game user emotion fluctuation situation, be convenient to the pouplarity understanding difference in functionality module; The classification of statistics network text according to demand, analyzes the game focus of different game user, obtains game potential user colony, and meanwhile, the leak in statistics game, is convenient to game developer and intuitively checks the various leaks that game user feeds back.
Correspondingly, the present invention also provides a kind of detection system of public sentiment of playing, and can realize all flow processs of the detection method of the game public sentiment in above-described embodiment.
See Fig. 2, be the structural representation of an embodiment of the detection system of game public sentiment provided by the invention, comprise:
Acquisition module 1, for the public sentiment data in collection network text;
Word-dividing mode 2, for carrying out participle to described public sentiment data, obtains the word in described network text;
Keyword acquisition module 3, for inquiring about the game keyword dictionary set up in advance, obtains the game keyword in described word;
Emotion value computing module 4, for according to the sentiment dictionary set up in advance, calculates the emotion value of described game keyword;
Web text classification module 5, for according to the game keyword in described network text, adopts Bayesian Classification Arithmetic, calculates the classification belonging to described network text; And,
Evaluation and test data obtaining module 6, for adding up the emotion value of described game keyword, the quantity of described network text and classification, obtains the evaluation and test information of game.
Concrete, described word-dividing mode 2, specifically for based on game word dictionary and stop words dictionary, adopts stammerer participle, carries out participle, obtain the word in described network text to described public sentiment data.
Further, the game keyword in described game keyword dictionary is comprised in described game word dictionary.
Concrete, described emotion value computing module 4 specifically comprises:
Judging unit, for inquiring about the sentiment dictionary set up in advance, judges whether the modification word of described game keyword is emotion word; Described sentiment dictionary is the database of the weights storing multiple emotion word and correspondence thereof, described modification word is the word between the beginning of the sentence of the network text at described game keyword and its place, or described modification word is a upper word of playing between keyword in the network text at described game keyword and its place;
Reading unit, during for judging described modification word as emotion word at described judging unit, reads the weights that described emotion word is corresponding; And,
Computing unit, for according to described weights, calculates the emotion value of described game keyword.
Further, the detection system of described game public sentiment also comprises:
Keyword classification module, for classifying to all game keywords, obtains game keyword categories; And,
Comprehensive emotion value computing module, for the emotion value according to game keyword each in described game keyword categories, calculates the comprehensive emotion value of described game keyword categories.
Concrete, described Web text classification module 5 specifically comprises:
Keyword classification unit, for based on the keyword classification training set generated in advance, classifies to the game keyword in described network text, obtains the training sample classification of described network text; And,
Web text classification unit, for adopting Bayesian Classification Arithmetic, calculate the probability that described network text belongs to each training sample classification respectively, the training sample classification of maximum probability is the classification of described network text.
Concrete, the computing formula of described Bayesian Classification Arithmetic is as follows:
P(Ci|d)＝P(d|Ci)P(Ci)L(d,Ci)
Wherein, P (Ci │ d) belongs to the probability of training sample classification Ci for network text d, for there is the probability of described network text d in described training sample classification Ci in P (d │ Ci), P (Ci) is the marginal probability of described training sample classification Ci, L (d, Ci) be the length factor of described network text d, Len (d) is the number of word in described network text d, avgLen (Ci) is the word mean number of network text in described training sample classification Ci, and k is the disturbance degree of word number to described training sample classification Ci.
The detection method of the game public sentiment that the embodiment of the present invention provides and system, can by the emotion value of keyword of playing in statistics public sentiment data, and the quantity of network text and classification, understands the Sentiment orientation of user to game, the evaluation and test information of Obtaining Accurate game; Based on game word dictionary and stop words dictionary, participle is carried out to public sentiment data, improve the accuracy rate to game proper noun participle, and avoid the redundancy keyword generation that has nothing to do in a large number; According to game keyword dictionary, obtain the game keyword in network text, the network text irrelevant with game content can be filtered out simultaneously, improve statistical efficiency and accuracy rate; Add up the emotion value of each game keyword, obtain the Sentiment orientation of user to game various piece; Word number in text Network Based, improves Bayesian Classification Arithmetic, and adopts the Bayesian Classification Arithmetic improved to classify to network text, improves accuracy rate and the recall rate of Web text classification.
One of ordinary skill in the art will appreciate that all or part of flow process realized in above-described embodiment method, that the hardware that can carry out instruction relevant by computer program has come, described program can be stored in a computer read/write memory medium, this program, when performing, can comprise the flow process of the embodiment as above-mentioned each side method.Wherein, described storage medium can be magnetic disc, CD, read-only store-memory body (Read-Only Memory, ROM) or random store-memory body (Random Access Memory, RAM) etc.
The above is the preferred embodiment of the present invention; it should be pointed out that for those skilled in the art, under the premise without departing from the principles of the invention; can also make some improvements and modifications, these improvements and modifications are also considered as protection scope of the present invention.
Claims (14)
1. a detection method for public sentiment of playing, is characterized in that, comprising:
Public sentiment data in collection network text;
Participle is carried out to described public sentiment data, obtains the word in described network text;
Inquire about the game keyword dictionary set up in advance, obtain the game keyword in described word;
According to the sentiment dictionary set up in advance, calculate the emotion value of described game keyword;
According to the game keyword in described network text, adopt Bayesian Classification Arithmetic, calculate the classification belonging to described network text;
Add up the emotion value of described game keyword, the quantity of described network text and classification, obtain the evaluation and test information of game.
2. the detection method of game public sentiment as claimed in claim 1, is characterized in that, describedly carries out participle to described public sentiment data, obtains the word in described network text, specifically comprises:
Based on game word dictionary and stop words dictionary, adopt stammerer participle, participle is carried out to described public sentiment data, obtains the word in described network text.
3. the detection method of game public sentiment as claimed in claim 2, is characterized in that, comprise the game keyword in described game keyword dictionary in described game word dictionary.
4. the detection method of game public sentiment as claimed in claim 1, is characterized in that, the sentiment dictionary that described basis is set up in advance, calculates the emotion value of described game keyword, specifically comprise:
Inquire about the sentiment dictionary set up in advance, judge whether the modification word of described game keyword is emotion word; Described sentiment dictionary is the database of the weights storing multiple emotion word and correspondence thereof, described modification word is the word between the beginning of the sentence of the network text at described game keyword and its place, or described modification word is a upper word of playing between keyword in the network text at described game keyword and its place;
If described modification word is emotion word, then read the weights that described emotion word is corresponding;
According to described weights, calculate the emotion value of described game keyword.
5. the detection method of game public sentiment as claimed in claim 1, is characterized in that, at the sentiment dictionary that described basis is set up in advance, after calculating the emotion value of described game keyword, also comprise:
All game keywords are classified, obtains game keyword categories;
According to the emotion value of game keyword each in described game keyword categories, calculate the comprehensive emotion value of described game keyword categories.
6. the detection method of game public sentiment as claimed in claim 1, is characterized in that, described according to the game keyword in described network text, adopts Bayesian Classification Arithmetic to classify to described network text, obtains the classification of described network text, specifically comprise:
Based on the keyword classification training set generated in advance, the game keyword in described network text is classified, obtain the training sample classification of described network text;
Adopt Bayesian Classification Arithmetic, calculate the probability that described network text belongs to each training sample classification respectively, the training sample classification of maximum probability is the classification of described network text.
7. the detection method of the game public sentiment as described in any one of claim 1 to 6, is characterized in that, the computing formula of described Bayesian Classification Arithmetic is as follows:
P(Ci|d)＝P(d|Ci)P(Ci)L(d,Ci)
Wherein, P (Ci │ d) belongs to the probability of training sample classification Ci for network text d, for there is the probability of described network text d in described training sample classification Ci in P (d │ Ci), P (Ci) is the marginal probability of described training sample classification Ci, L (d, Ci) be the length factor of described network text d, Len (d) is the number of word in described network text d, avgLen (Ci) is the word mean number of network text in described training sample classification Ci, and k is the disturbance degree of word number to described training sample classification Ci.
8. a detection system for public sentiment of playing, is characterized in that, comprising:
Acquisition module, for the public sentiment data in collection network text;
Word-dividing mode, for carrying out participle to described public sentiment data, obtains the word in described network text;
Keyword acquisition module, for inquiring about the game keyword dictionary set up in advance, obtains the game keyword in described word;
Emotion value computing module, for according to the sentiment dictionary set up in advance, calculates the emotion value of described game keyword;
Web text classification module, for according to the game keyword in described network text, adopts Bayesian Classification Arithmetic, calculates the classification belonging to described network text; And,
Evaluation and test data obtaining module, for adding up the emotion value of described game keyword, the quantity of described network text and classification, obtains the evaluation and test information of game.
9. the detection system of game public sentiment as claimed in claim 8, it is characterized in that, described word-dividing mode, specifically for based on game word dictionary and stop words dictionary, adopts stammerer participle, participle is carried out to described public sentiment data, obtains the word in described network text.
10. the detection system of game public sentiment as claimed in claim 9, is characterized in that, comprise the game keyword in described game keyword dictionary in described game word dictionary.
The detection system of 11. game public sentiments as claimed in claim 8, it is characterized in that, described emotion value computing module specifically comprises:
Judging unit, for inquiring about the sentiment dictionary set up in advance, judges whether the modification word of described game keyword is emotion word; Described sentiment dictionary is the database of the weights storing multiple emotion word and correspondence thereof, described modification word is the word between the beginning of the sentence of the network text at described game keyword and its place, or described modification word is a upper word of playing between keyword in the network text at described game keyword and its place;
Reading unit, during for judging described modification word as emotion word at described judging unit, reads the weights that described emotion word is corresponding; And,
Computing unit, for according to described weights, calculates the emotion value of described game keyword.
The detection system of 12. game public sentiments as claimed in claim 8, it is characterized in that, the detection system of described game public sentiment also comprises:
Keyword classification module, for classifying to all game keywords, obtains game keyword categories; And,
Comprehensive emotion value computing module, for the emotion value according to game keyword each in described game keyword categories, calculates the comprehensive emotion value of described game keyword categories.
The detection system of 13. game public sentiments as claimed in claim 8, it is characterized in that, described Web text classification module specifically comprises:
Keyword classification unit, for based on the keyword classification training set generated in advance, classifies to the game keyword in described network text, obtains the training sample classification of described network text; And,
Web text classification unit, for adopting Bayesian Classification Arithmetic, calculate the probability that described network text belongs to each training sample classification respectively, the training sample classification of maximum probability is the classification of described network text.
The detection system of 14. game public sentiments as described in any one of claim 8 to 13, it is characterized in that, the computing formula of described Bayesian Classification Arithmetic is as follows:
P(Ci|d)＝P(d|Ci)P(Ci)L(d,Ci)
Wherein, P (Ci │ d) belongs to the probability of training sample classification Ci for network text d, for there is the probability of described network text d in described training sample classification Ci in P (d │ Ci), P (Ci) is the marginal probability of described training sample classification Ci, L (d, Ci) be the length factor of described network text d, Len (d) is the number of word in described network text d, avgLen (Ci) is the word mean number of network text in described training sample classification Ci, and k is the disturbance degree of word number to described training sample classification Ci.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410805964.3A CN104572877A (en) | 2014-12-22 | 2014-12-22 | Detection method and detection system of game public opinion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410805964.3A CN104572877A (en) | 2014-12-22 | 2014-12-22 | Detection method and detection system of game public opinion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104572877A true CN104572877A (en) | 2015-04-29 |
Family
ID=53088939
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410805964.3A Pending CN104572877A (en) | 2014-12-22 | 2014-12-22 | Detection method and detection system of game public opinion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104572877A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105824959A (en) * | 2016-03-31 | 2016-08-03 | 首都信息发展股份有限公司 | Public opinion monitoring method and system |
| CN106250363A (en) * | 2016-07-15 | 2016-12-21 | 合肥指南针电子科技有限责任公司 | A kind of public sentiment monitoring analysis method |
| CN107645559A (en) * | 2017-09-30 | 2018-01-30 | 广东美的制冷设备有限公司 | Household electrical appliances information-pushing method, server, mobile terminal and storage medium |
| CN108228612A (en) * | 2016-12-14 | 2018-06-29 | 北京国双科技有限公司 | A kind of method and device for extracting network event keyword and mood tendency |
| CN110489653A (en) * | 2019-08-23 | 2019-11-22 | 北京金堤科技有限公司 | Public feelings information querying method and device, system, electronic equipment, storage medium |
| CN110600033A (en) * | 2019-08-26 | 2019-12-20 | 北京大米科技有限公司 | Learning condition evaluation method and device, storage medium and electronic equipment |
| CN112380341A (en) * | 2020-11-09 | 2021-02-19 | 恒瑞通（福建）信息技术有限公司 | Public opinion analysis method and terminal based on administrative service center |
| CN112749269A (en) * | 2019-10-31 | 2021-05-04 | 北京国双科技有限公司 | Entity public opinion calculation method and system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101609459A (en) * | 2009-07-21 | 2009-12-23 | 北京大学 | A kind of extraction system of affective characteristic words |
| CN102708164A (en) * | 2012-04-26 | 2012-10-03 | 苏州大学 | Method and system for calculating movie expectation |
| CN102789498A (en) * | 2012-07-16 | 2012-11-21 | 钱钢 | Method and system for carrying out sentiment classification on Chinese comment text on basis of ensemble learning |
| CN102890707A (en) * | 2012-08-28 | 2013-01-23 | 华南理工大学 | System for mining emotional tendencies of brief network comments based on conditional random field |
| CN103150367A (en) * | 2013-03-07 | 2013-06-12 | 宁波成电泰克电子信息技术发展有限公司 | Method for analyzing emotional tendency of Chinese microblogs |
| CN103399916A (en) * | 2013-07-31 | 2013-11-20 | 清华大学 | Internet comment and opinion mining method and system on basis of product features |
| CN103425777A (en) * | 2013-08-15 | 2013-12-04 | 北京大学 | Intelligent short message classification and searching method based on improved Bayesian classification |
-
2014
- 2014-12-22 CN CN201410805964.3A patent/CN104572877A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101609459A (en) * | 2009-07-21 | 2009-12-23 | 北京大学 | A kind of extraction system of affective characteristic words |
| CN102708164A (en) * | 2012-04-26 | 2012-10-03 | 苏州大学 | Method and system for calculating movie expectation |
| CN102789498A (en) * | 2012-07-16 | 2012-11-21 | 钱钢 | Method and system for carrying out sentiment classification on Chinese comment text on basis of ensemble learning |
| CN102890707A (en) * | 2012-08-28 | 2013-01-23 | 华南理工大学 | System for mining emotional tendencies of brief network comments based on conditional random field |
| CN103150367A (en) * | 2013-03-07 | 2013-06-12 | 宁波成电泰克电子信息技术发展有限公司 | Method for analyzing emotional tendency of Chinese microblogs |
| CN103399916A (en) * | 2013-07-31 | 2013-11-20 | 清华大学 | Internet comment and opinion mining method and system on basis of product features |
| CN103425777A (en) * | 2013-08-15 | 2013-12-04 | 北京大学 | Intelligent short message classification and searching method based on improved Bayesian classification |
Non-Patent Citations (1)
| Title |
|---|
| 阳锋: ""基于主题模型的情感搜索引擎的研究与实现"", 《中国优秀硕士学位论文全文数据库（电子期刊）》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105824959A (en) * | 2016-03-31 | 2016-08-03 | 首都信息发展股份有限公司 | Public opinion monitoring method and system |
| CN105824959B (en) * | 2016-03-31 | 2021-09-10 | 首都信息发展股份有限公司 | Public opinion monitoring method and system |
| CN106250363A (en) * | 2016-07-15 | 2016-12-21 | 合肥指南针电子科技有限责任公司 | A kind of public sentiment monitoring analysis method |
| CN108228612A (en) * | 2016-12-14 | 2018-06-29 | 北京国双科技有限公司 | A kind of method and device for extracting network event keyword and mood tendency |
| CN108228612B (en) * | 2016-12-14 | 2022-03-18 | 北京国双科技有限公司 | Method and device for extracting network event keywords and emotional tendency |
| CN107645559A (en) * | 2017-09-30 | 2018-01-30 | 广东美的制冷设备有限公司 | Household electrical appliances information-pushing method, server, mobile terminal and storage medium |
| CN107645559B (en) * | 2017-09-30 | 2020-10-09 | 广东美的制冷设备有限公司 | Household appliance information pushing method, server, mobile terminal and storage medium |
| CN110489653A (en) * | 2019-08-23 | 2019-11-22 | 北京金堤科技有限公司 | Public feelings information querying method and device, system, electronic equipment, storage medium |
| CN110600033A (en) * | 2019-08-26 | 2019-12-20 | 北京大米科技有限公司 | Learning condition evaluation method and device, storage medium and electronic equipment |
| CN112749269A (en) * | 2019-10-31 | 2021-05-04 | 北京国双科技有限公司 | Entity public opinion calculation method and system |
| CN112749269B (en) * | 2019-10-31 | 2024-06-21 | 北京国双科技有限公司 | Entity public opinion calculation method and system |
| CN112380341A (en) * | 2020-11-09 | 2021-02-19 | 恒瑞通（福建）信息技术有限公司 | Public opinion analysis method and terminal based on administrative service center |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104572877A (en) | Detection method and detection system of game public opinion | |
| West et al. | Human wayfinding in information networks | |
| CN102663139B (en) | Method and system for constructing emotional dictionary | |
| CN106980692A (en) | A kind of influence power computational methods based on microblogging particular event | |
| CN103177090B (en) | A kind of topic detection method and device based on big data | |
| Liao et al. | Evaluating the effectiveness of search task trails | |
| CN103077190A (en) | Hot event ranking method based on order learning technology | |
| CN106940732A (en) | A kind of doubtful waterborne troops towards microblogging finds method | |
| CN106126582A (en) | Recommend method and device | |
| CN104967587B (en) | A kind of recognition methods of malice account and device | |
| CN102929873A (en) | Method and device for extracting searching value terms based on context search | |
| CN106202372A (en) | A kind of method of network text information emotional semantic classification | |
| CN102831193A (en) | Topic detecting device and topic detecting method based on distributed multistage cluster | |
| CN104268197A (en) | Industry comment data fine grain sentiment analysis method | |
| CN105354216B (en) | A kind of Chinese microblog topic information processing method | |
| CN103699626A (en) | Method and system for analysing individual emotion tendency of microblog user | |
| CN107291886A (en) | A kind of microblog topic detecting method and system based on incremental clustering algorithm | |
| CN103177024A (en) | Method and device of topic information show | |
| CN112104642B (en) | Abnormal account number determination method and related device | |
| CN104102658B (en) | Content of text method for digging and device | |
| CN103559174B (en) | Semantic emotion classification characteristic value extraction and system | |
| CN103440235A (en) | Method and device for identifying text emotion types based on cognitive structure model | |
| CN104376010A (en) | User recommendation method and user recommendation device | |
| CN106126613A (en) | One composition of digressing from the subject determines method and device | |
| CN102880600A (en) | Word semantic tendency prediction method based on universal knowledge network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150429 |