CN102236722A - Method and system for generating user comment summaries based on triples - Google Patents

Method and system for generating user comment summaries based on triples Download PDF

Info

Publication number
CN102236722A
CN102236722A CN2011102366837A CN201110236683A CN102236722A CN 102236722 A CN102236722 A CN 102236722A CN 2011102366837 A CN2011102366837 A CN 2011102366837A CN 201110236683 A CN201110236683 A CN 201110236683A CN 102236722 A CN102236722 A CN 102236722A
Authority
CN
China
Prior art keywords
tlv triple
feature
decision
making
comment
Prior art date
Application number
CN2011102366837A
Other languages
Chinese (zh)
Other versions
CN102236722B (en
Inventor
石忠民
徐亚波
杜伟夫
Original Assignee
广州索答信息科技有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 广州索答信息科技有限公司 filed Critical 广州索答信息科技有限公司
Priority to CN201110236683.7A priority Critical patent/CN102236722B/en
Publication of CN102236722A publication Critical patent/CN102236722A/en
Application granted granted Critical
Publication of CN102236722B publication Critical patent/CN102236722B/en

Links

Abstract

The invention discloses a method and a system for generating user comment summaries based on triples. The method comprises the following steps of: establishing a feature word bank, a mapping word list and an emotional word bank of objects, and constructing a feature tree according to the feature word bank; grasping a user comment webpage; receiving user comments; processing each user comment one by one and generating own comment triple-based comment summary; summing up and integrating the comment triples of all the user comments to generate decision triples; calculating the number of decision triples in which the feature words and the emotional words are the same in polarity; and extracting all the decision triples to generate a decision summary. With the method or the system of the present invention, the comment summary is generated for each user comment so that the user can check and refer to this information; and all the comment triples are summed up and integrated to generate the decision triples having the directing significance; furthermore, all the decision triples are extracted to generate the decision summary which is capable of reflecting the overall assessment results and having the function of assisting decision; therefore, the user can be helped to make a correct decision quickly.

Description

A kind of generation method and system of making a summary based on the user comment of tlv triple

Technical field

The present invention relates to computerized information digging technology field, relate in particular to a kind of generation method and system of making a summary, be mainly used in and from a large number of users comment of object, generate a decision-making summary that can objectively respond the overall assessment result of all user comments based on the user comment of tlv triple.

Background technology

At present, along with popularizing of internet, the user wishes to go to understand the comment that other users deliver consumer objects by the internet before consumption, determine with this whether consumer objects is worth oneself going consumption, this object can be businessman or product, it also can be service, think certain restaurant or market consumption such as the user, the advertisement information of only seeing this restaurant or market is not enough, because these advertisement informations are difficult to objectively describe its real product quality and service level, the user would like to know in this restaurant naturally or how other users of market post-consumer estimate it.Yet, when at the user comment of object very many time, the user is difficult to know how many front evaluation and negative evaluations of certain feature of oneself paying special attention to of this object respectively account for from large-scale user comment, be difficult to also learn that the total result of all user comments is that the in the majority or negative evaluation of positive evaluation is in the majority, think that such as the user restaurant has a meal, pay special attention to the food and the environment in this restaurant, but the distribution of user comment in all user comments that relates to food and environment is irregular, the user wants to check that all user comments that relate to food and environment must finish watching whole user comments one by one, oneself also will do statistics to front evaluation and negative evaluation, this is the power that consumes again consuming time obviously, extremely inconvenient, and cost big cost like this is finished watching whole user comments, what know also only is the evaluation result of food and these two features of environment, wonder the evaluation result of further feature and the overall assessment result of all user comments, its workload is unthinkable.And the literal length of a user comment has the long weak point that has, and wherein the information that the user paid close attention to is the feature of object and the emotion speech of describing feature, and out of Memory all is useless, but the user can not only check the own information of being paid close attention to when checking.

In sum, the user has following two significant problems when checking the user comment of object at present:

1. be flooded with a large amount of garbages in the user comment, lose time when causing checking;

2. how many front evaluation and the negative evaluations that can not directly check feature respectively account for, and can not learn the overall assessment result of all user comments, though therefore large-scale user comment is arranged as a reference, assisted user is made correct decisions apace intuitively.

Summary of the invention

At the deficiencies in the prior art, fundamental purpose of the present invention is intended to provide a kind of generation method of making a summary based on the user comment of tlv triple.

Another object of the present invention provides a kind of generation system of making a summary based on the user comment of tlv triple.

The present invention adopts following technical scheme for achieving the above object:

A kind of generation method of making a summary based on the user comment of tlv triple comprises:

Step 1. is set up feature dictionary, mapping vocabulary and the emotion dictionary of object, and according to the characteristics tree of the feature construction object in the feature dictionary, wherein, mapping speech in the mapping vocabulary is corresponding with the Feature Mapping in the feature dictionary, the emotion dictionary comprises positive emotion dictionary and negative emotion dictionary, the root node on characteristics tree top is an object, each layer leaf node under the root node is the feature of object, and following one deck leaf node is the subcharacter of the last layer leaf node of correspondence, and the last layer leaf node is father's feature of following one deck leaf node of correspondence;

Step 2. is the directed user comment webpage that grasps object from the internet;

Step 3. receives all user comments of object in the user comment webpage;

Step 4. is carried out following processing one by one to each user comment, generates comment summary separately:

Step 41. extracts the feature of object according to feature dictionary and mapping vocabulary;

Step 42. is according to emotion dictionary identification emotion speech;

Step 43. collocation feature and emotion speech generate the comment tlv triple based on object, feature, emotion speech;

Step 44. extracts the comment summary that the comment tlv triple generates this user comment;

And this method also comprises:

Step 5. is concluded the comment tlv triple of integrating all user comments, the feature that will be the comment tlv triple of feature with the leaf node below the characteristics tree ground floor leaf node sums up in the point that on its corresponding ground floor leaf node that generating all is the decision-making tlv triple of feature with characteristics tree ground floor leaf node;

The quantity of the decision-making tlv triple that step 6. calculated characteristics is identical with emotion speech polarity, if quantity equals 1, with this decision-making tlv triple and incompatible this decision-making tlv triple of representing of sets of numbers, if quantity is greater than 1, with any one decision-making tlv triple wherein with sets of numbers is incompatible represents that these have the decision-making tlv triple of same characteristic features and emotion speech polarity;

Step 7. extracts the decision-making summary that all decision-making tlv triple generate all user comments with the representation of decision-making tlv triple and number combinations.

As a kind of preferred version, described step 41 comprises:

Step 411. is divided into sentence with user comment;

Each speech in the step 412. traversal sentence, judge whether it occurs in the feature dictionary, if, extract as feature, if do not appear in the feature dictionary but appear in the mapping vocabulary, extract in the feature dictionary to come out as feature with this speech mapping characteristic of correspondence.

As a kind of preferred version, described step 42 comprises:

Step 421. is divided into sentence with user comment;

Each speech in the step 422. traversal sentence extracts the speech that appears in the emotion dictionary as the emotion speech;

Step 423. is judged the polarity of the emotion speech that extracts according to the polarity of emotion dictionary.

As a kind of preferred version, described step 43 comprises:

Step 431. is extracted the feature templates of tlv triple from training sample;

Step 432. uses the svm classifier method according to sorter of feature templates training;

Step 433. utilizes syntax rule that feature and emotion speech are made up, and generates tlv triple;

Step 434. utilizes sorter that feature and emotion speech are arranged in pairs or groups, and generates tlv triple;

Step 435. utilizes candidate's tlv triple set pair of artificial mark to be filtered by all tlv triple that syntax rule and sorter generate, and removes feature and emotion speech irrational tlv triple of arranging in pairs or groups, and obtains commenting on tlv triple.

As a kind of preferred version, described step 6 also comprises the quantity of calculating positive decision-making tlv triple and the quantity of negative decision-making tlv triple, and described step 7 also comprises the content of the quantity of the quantity that extracts this front decision-making tlv triple and negative decision-making tlv triple as described decision-making summary.

A kind of generation system of making a summary based on the user comment of tlv triple comprises:

Pretreatment unit, be used to set up feature dictionary, mapping vocabulary and the emotion dictionary of object, and according to the characteristics tree of the feature construction object in the feature dictionary, wherein, mapping speech in the mapping vocabulary is corresponding with the Feature Mapping in the feature dictionary, the emotion dictionary comprises positive emotion dictionary and negative emotion dictionary, the root node on characteristics tree top is an object, each layer leaf node under the root node is the feature of object, and following one deck leaf node is the subcharacter of the last layer leaf node of correspondence, and the last layer leaf node is father's feature of following one deck leaf node of correspondence;

The reptile device is used for the directed user comment webpage that grasps object from the internet;

Receiving trap is used for receiving all user comments of user comment webpage object;

Treating apparatus is used for each user comment is handled one by one, generates comment summary separately, and this treating apparatus comprises:

The feature extraction device is used for the feature according to feature dictionary and mapping vocabulary extraction object;

Emotion speech recognition device is used for according to emotion dictionary identification emotion speech;

Comment tlv triple generating apparatus, be used to arrange in pairs or groups feature and emotion speech generate the comment tlv triple based on object, feature, emotion speech;

Comment summary generating apparatus is used to extract the comment summary that the comment tlv triple generates this user comment;

And this system also comprises:

Decision-making tlv triple generating apparatus, be used to conclude the comment tlv triple of integrating all user comments, the feature that will be the comment tlv triple of feature with the leaf node below the characteristics tree ground floor leaf node sums up in the point that on its corresponding ground floor leaf node that generating all is the decision-making tlv triple of feature with characteristics tree ground floor leaf node;

Calculation element, the quantity that is used for the calculated characteristics decision-making tlv triple identical with emotion speech polarity, if quantity equals 1, with this decision-making tlv triple and incompatible this decision-making tlv triple of representing of sets of numbers, if quantity is greater than 1, with any one decision-making tlv triple wherein with sets of numbers is incompatible represents that these have the decision-making tlv triple of same characteristic features and emotion speech polarity;

Decision-making summary generating apparatus is used for extracting the decision-making summary that all decision-making tlv triple generate all user comments with the representation of decision-making tlv triple and number combinations.

As a kind of preferred version, described feature extraction device comprises:

The device that user comment is divided into sentence;

Each speech in the traversal sentence, judge whether it occurs in the feature dictionary, if, extract as feature, if do not appear in the feature dictionary but appear in the mapping vocabulary, extract in the feature dictionary to come out as the device of feature with this speech mapping characteristic of correspondence.

As a kind of preferred version, described emotion speech recognition device comprises:

The device that user comment is divided into sentence;

Traversal each speech in the sentence extracts device as the emotion speech with appearing at speech in the emotion dictionary;

Judge the device of the polarity of the emotion speech that extracts according to the polarity of emotion dictionary.

As a kind of preferred version, described comment tlv triple generating apparatus comprises:

From training sample, extract the device of the feature templates of tlv triple;

Use the device of svm classifier method according to a sorter of feature templates training;

Utilize syntax rule that feature and emotion speech are made up, generate the device of tlv triple;

Utilize sorter that feature and emotion speech are arranged in pairs or groups, generate the device of tlv triple;

Utilize candidate's tlv triple set pair of artificial mark to filter, remove feature and emotion speech irrational tlv triple of arranging in pairs or groups, obtain commenting on the device of tlv triple by all tlv triple that syntax rule and sorter generate.

As a kind of preferred version, described calculation element also is used to calculate the quantity of positive decision-making tlv triple and the quantity of negative decision-making tlv triple, and described decision-making summary generating apparatus also is used to extract the content of the quantity of the quantity of this front decision-making tlv triple and negative decision-making tlv triple as described decision-making summary.

A kind of generation method and system of making a summary set forth in the present invention based on the user comment of tlv triple, its beneficial effect is: utilize this method or system, with the feature in each user comment, the emotion speech extracts generation based on object, feature, the comment tlv triple of emotion speech, for each user comment generates its comment summary based on the comment tlv triple, realized the information that the user paid close attention in the user comment is extracted succinctly summary info intuitively of formation separately, so that the user checks reference, and, conclude integration by commenting on tlv triple, generation has the decision-making tlv triple of directive significance, and extract all the decision-making tlv triple with the representation of decision-making tlv triple and number combinations and generate and to reflect the overall assessment result, decision-making summary with decision-making booster action, in the decision-making summary, the user can check directly that the front of own feature of being paid close attention to and further feature is estimated and how many negative evaluations respectively accounts for, also can know the overall assessment result of all user comments, thereby assisted user is made correct decisions apace.

Description of drawings

Fig. 1 is the schematic flow sheet of the generation method of a kind of user comment summary based on tlv triple of the present invention.

Fig. 2 is the configuration diagram of a characteristics tree.

Fig. 3 is the diagrammatic sketch of a user comment and comment summary thereof.

Fig. 4 is a decision-making summary diagrammatic sketch.

Embodiment

Come the present invention is further described below in conjunction with accompanying drawing and specific embodiment.

Please refer to shown in Figure 1ly, it has demonstrated the overall procedure of the generation method of a kind of user comment summary based on tlv triple of the present invention.In step (1), set up feature dictionary, mapping vocabulary and the emotion dictionary of object, and according to the characteristics tree of the feature construction object in the feature dictionary, wherein:

Object can be product, businessman or service, and the feature dictionary is the set of the speech that can be used as characteristics of objects collected from extensive language material.The foundation of feature dictionary can use the method based on statistics to realize that the specific implementation process can be: at first, collect a seed characteristics dictionary that has comprised all nouns from extensive language material; Then, the frequency that all nouns occur in extensive language material in the statistics seed characteristics dictionary; Then, the noun that the frequency of occurrences is lower than pre-set threshold value removes as stop words, generates the initial characteristics dictionary; At last, the speech in the initial characteristics dictionary is filtered, generate final feature dictionary.

Mapping speech in the mapping vocabulary is corresponding with the Feature Mapping in the feature dictionary, the purpose of setting up the mapping vocabulary is the potential feature that may exist in the user comment in order to excavate, the definition of potential feature is relative explicit features, if the feature in the feature dictionary in user comment, occurred then this feature is an explicit features in this user comment, and because the dirigibility of Chinese and user's expression problem, though the user may estimate certain feature of object when making comments, but do not write out this feature in the literal, then this feature is potential feature in this user comment, such as being certain user comment in certain restaurant at object, the user has write " eating on the contrary nice; be exactly too expensive " in comment, just do not write out feature in the words, but " eat " and but implied " food " this feature in this verb, therefore here " eating " is the mapping speech, and " food " is the potential feature corresponding with this mapping speech.The mapping vocabulary is exactly the set of the mapping speech that has comprised potential feature collected from extensive language material, how the mapping speech of mapping in the vocabulary selects the feature in the feature dictionary can be by the PMI(Point-wise Mutual Information between calculated characteristics and the mapping speech as potential feature, the pointwise mutual information) determines, computing formula is: PMI(f, d)=hits(f, d)/hits(f) hits(d), wherein f is a feature, d is the mapping speech, high more this feature that shows of PMI value is just big more as the possibility of the potential feature of this mapping speech, therefore generally is the corresponding relation that the highest mapping speech of collocation PMI value and feature are set up mapping vocabulary and feature dictionary.

The emotion dictionary comprises positive emotion dictionary and negative emotion dictionary, the emotion dictionary is the set of an emotion speech that has obvious emotion tendency of collecting from extensive language material, the emotion speech of two kinds of feeling polarities only collected in the emotion dictionary, a kind of is positive, for example " good ", " satisfaction " is exactly the emotion speech in two fronts, it is negative also having a kind of, for example " disappointing " is exactly a negative emotion speech, because these two kinds of diametrically opposite feeling polarities can provide reference value for the user, and relatively more neutral emotion speech meaning and little concerning the user, the foundation of emotion dictionary can be used the method based on statistics.

The root node on characteristics tree top is an object, and each layer leaf node under the root node is the feature of object, and one deck leaf node is the subcharacter of the last layer leaf node of correspondence down, and the last layer leaf node is father's feature of following one deck leaf node of correspondence.Characteristics tree has defined the feature of current object and the relation between the feature, this relation is with different levels tree structure, in characteristics tree, get over the extensive notion of node on upper strata, father's feature is the summary to the attribute of its all subcharacters, subcharacter is from different perspectives to the refinement of his father's feature, forms relations on an equal basis between all subcharacters of same father's feature.For ease of explanation, a restaurant with catering field is an example, please refer to shown in Figure 2, it is the sketch of the characteristics tree of object with " restaurant " that Fig. 2 is one, what be in characteristics tree top root node among Fig. 2 is " restaurant " this object, " food " in the ground floor leaf node, " service ", " cost performance ", " environment " is to summarize the feature of " restaurant " attribute, " nutrition " in the second layer leaf node is the subcharacter of " food ", " atmosphere ", " tableware " is the subcharacter of " environment " with " finishing ", " shop front ", " tone ", " style " is again the subcharacter of " finishing ", the structure of characteristics tree can use in conjunction with statistical machine study sorts out integration with rule-based method to the feature in the feature dictionary, take out different concept hierarchies, generate required characteristics tree.

Proceed to step (2), the directed user comment webpage that grasps object from the internet.This relates to web crawlers, gets the Internet resources relevant with theme in order to climb efficiently, and general adoptable climbing got strategy and related algorithm has: based on the heuristic of word content; Method based on the super chain figure evaluation of web; Based on the sorter forecast method; Other focused crawl methods.

Proceed to step (3), receive all user comments of object in the user comment webpage.This relates to the web page text information extraction technique, can adopt web page text information extraction based on dispenser, Web page text extracting based on statistics, web page blocks analysis based on vision, technology such as Web page text extracting based on data mining thought realize, also can adopt a kind of scheme stage by stage: the phase one remains the technology path based on dispenser, yet with common based on the information extraction technique of dispenser different be, for the dispenser part, every dispenser decimation rule all is configurableization, utilize the xml analytic technique to realize, more particularly, decimation rule is based on the xpath inquiry, make extraction very convenient flexibly, and for the specifically generation of each xpath, then be the mode of utilization autopager browser plug-in, the auxiliary generation has certain semi-automatic characteristics; The method of subordinate phase utilization machine learning, for structured message web webpage to be extracted, according to the characteristics of structured message to be extracted, the utilization heuritic approach is discerned its corresponding xpath automatically, and generate respective x path configuration file, realize the automatic extraction of wrapper rule.

Proceed to step (4), each user comment is handled one by one, extract the information that the user pays close attention in the user comment: feature, the emotion speech, and feature and emotion speech reasonably arranged in pairs or groups according to certain rule, generate separately based on object, feature, the comment tlv triple of emotion speech, comment tlv triple generation comment summary separately by each user comment, the comment tlv triple has reflected the viewpoint of this user comment, the comment summary then succinctly illustrates the evaluation result that this user comment is expressed intuitively, on the UI interface, the comment summary can be presented at the right side of user comment, please refer to shown in Figure 3, it has demonstrated a user comment and its comment summary, and among Fig. 3, user " abc " delivered a comment of estimating certain restaurant on May 1st, 2009, the comment content be " the service here is well; and popularity is also many, and environment is also good, and price is relatively to calculate! I will come twice at a week, everybody consumption that can come here more! Wrap you and come on an impulse, return in high spirits."; generated its comment summary according to this user comment; the content of comment summary is " the pretty good popularity of environment is served good price more and calculated "; four comment tlv triple are wherein arranged; be respectively<the restaurant environment is pretty good 〉;<restaurant popularity is many 〉,<restaurant is served 〉,<restaurant price is calculated 〉, certainly, when on the UI interface, showing, because the object of all comment tlv triple and decision-making tlv triple all is identical, therefore can be with Objects hide, make the interface more succinct, four comment tlv triple of comment summary have just been hidden its common object " restaurant " among Fig. 3, as can be seen from Figure 3, the information that the user paid close attention in this user comment has been extracted has separately formed the comment summary that comprises four comment tlv triple, is convenient to very much the evaluation that the user knows that this user comment is made.The detailed process that this step (4) is handled each user comment is: step (41), feature according to feature dictionary and mapping vocabulary extraction object, feature is the attribute of object, the basal conditions of energy reflection object, for example, for " restaurant " this object, its feature just has " food ", " service ", " cost performance ", " environment " etc.; Step (42), according to emotion dictionary identification emotion speech, the emotion speech is to be used for expressing the tendentious word of viewpoint, have tangible subjectivity, the user estimates the quality of certain feature with it, for example, for the food and drink comment, speech such as " height ", " low ", " satisfaction ", " disappointing " are exactly some common emotion speech; Step (43), collocation feature and emotion speech generate the comment tlv triple based on object, feature, emotion speech; Step (44) extracts the comment summary that the comment tlv triple generates this user comment.Wherein, when carrying out step (41), can at first user comment be divided into sentence by punctuation mark, then travel through each speech in the sentence, judge whether it occurs in the feature dictionary, if, it is extracted as explicit features, if do not appear in the feature dictionary but appear in the mapping vocabulary, then will shine upon speech mapping characteristic of correspondence and come out as potential feature extraction with this, the mapping speech generally is emotion speech or verb; When carrying out step (42), can by punctuation mark user comment be divided into sentence earlier equally, then travel through each speech in the sentence, the speech that appears in the emotion dictionary is extracted as the emotion speech, judge the polarity of the emotion speech that extracts then according to the polarity of emotion dictionary; When carrying out step (43), can use the method for machine learning, and the fusion syntactic feature judges that to the relation between feature speech and the emotion speech concrete grammar is: at first, from training sample, extract the feature templates of tlv triple; Then, use the svm classifier method according to sorter of feature templates training; Then, utilize syntax rule that feature and emotion speech are made up, generate tlv triple, utilize sorter that feature and emotion speech are arranged in pairs or groups, generate tlv triple; At last, utilize candidate's tlv triple set pair of artificial mark to filter, remove feature and emotion speech irrational tlv triple of arranging in pairs or groups, obtain commenting on tlv triple by all tlv triple that syntax rule and sorter generate.Candidate's tlv triple collection be the current object that pre-defines the reasonable triplet sets that might exist, do not appear at tlv triple in candidate's triplet sets and all belong to feature and emotion speech irrational tlv triple of arranging in pairs or groups, but direct filtration is fallen, such as feature is " environment ", the emotion speech is " expensive ", then obviously be a unreasonable collocation, can not be present in candidate's triplet sets.

Proceed to step (5), conclude the comment tlv triple of integrating all user comments, the feature that will be the comment tlv triple of feature with the leaf node below the characteristics tree ground floor leaf node sums up in the point that on its corresponding ground floor leaf node that generating all is the decision-making tlv triple of feature with characteristics tree ground floor leaf node.For an object, may have many consumers it is commented on, and the viewpoint of different comments may be identical, also may be different, in addition opposite fully.Correspondingly, different comment tlv triple may be to the existing front of evaluation of same feature, also have negative,, the result who causes do not know that or existing comment is positive in the majority negative in the majority to the evaluation of these features after having seen these comment triplet information actually being the user, moreover the user disperses relatively to the evaluation of a certain feature, such as, concerning " environment " this feature, the user may comment on from all angles such as " finishing ", " health ", " atmosphere ".These characteristic evaluatings that disperse relatively are unfavorable for that the user makes intuitive judgment to a certain feature rapidly and accurately.The decision-making tlv triple is exactly to recognize the evaluation situation of the feature that can summarize object properties for assisted user, and the feature of characteristics tree ground floor leaf node is undoubtedly the most representative feature of object, therefore select the feature of ground floor leaf node as the decision-making tlv triple for use, other all features sum up in the point that all these features get on.Such as " environment " is a ground floor leaf node, be used as the feature of decision-making tlv triple, the feature of all relevant environment in the comment tlv triple, to sum up in the point that all " environment " this feature gets on, thereby generate the decision-making tlv triple, for example one to as if " restaurant ", feature is " atmosphere ", the emotion speech is that the comment tlv triple<restaurant atmosphere of " good " is good 〉, its feature " atmosphere " is summed up in the point that just to have generated one<restaurant environment after " environment " good〉the decision-making tlv triple, certainly, if original is the comment tlv triple of feature with characteristics tree ground floor leaf node exactly, then generated the decision-making tlv triple identical with this comment tlv triple, the same with aforementioned comment tlv triple, all the decision-making tlv triple with same object also can not demonstrate object on the UI interface.

Proceed to step (6), the quantity of the decision-making tlv triple that calculated characteristics is identical with emotion speech polarity is if quantity is greater than 1, with any one decision-making tlv triple wherein with sets of numbers is incompatible represents that these have the decision-making tlv triple of same characteristic features and emotion speech polarity.Certainly, if quantity equals 1, illustrating does not have other decision-making tlv triple and this decision-making tlv triple to have identical feature and emotion speech polarity, and the decision-making tlv triple with combination of numbers does not just have other selection so, can only be with this make a strategic decision tlv triple and incompatible this decision-making tlv triple of representing of sets of numbers.In this step, also can calculate the quantity of positive decision-making tlv triple and the quantity of negative decision-making tlv triple.

Proceed to step (7), representation with decision-making tlv triple and number combinations extracts the decision-making summary that all decision-making tlv triple generate all user comments, also can extract the content of the quantity of the quantity of positive decision-making tlv triple and negative decision-making tlv triple as the decision-making summary.In the decision-making summary, the user can check directly that the front of own feature of being paid close attention to and further feature is estimated and how many negative evaluations respectively accounts for, and also can know the overall assessment result of all user comments, thereby assisted user is made correct decisions apace.As shown in Figure 4, it has demonstrated the decision-making summary of all user comments in " A restaurant ", be how many front decision-making tlv triple and negative decision-making tlv triple of feature respectively have with " food ", " environment ", " service ", " taste ", " cost performance " as can be seen from Figure 4, also show the quantity of all positive decision-making tlv triple and negative decision-making tlv triple, this decision-making summary provides very useful reference information for the user makes high-speed decision.

Be pointed out that, more than repeatedly give an example with catering field, just for the ease of understanding the present invention, be not to limit application of the present invention, the present invention can be widely used in the field of any relevant product, businessman, service.

The present invention needs user comment is carried out participle when extracting feature and emotion speech, and the quality of participle performance has very important influence to the generation of tlv triple.The present invention adopts Hidden Markov Model (HMM) (HMM) to carry out participle and part-of-speech tagging, and use is carried out participle based on the method for Delimiter and obtained.And participle performance of the present invention is evaluated and tested, and evaluation method common in the natural language processing is continued to use in evaluation and test:

Accuracy rate: P=C3/C2;

Recall rate: R=C3/C1;

F value: F=2*P*R/ (P+R);

Wherein, C1 is the number of the speech of reality in the language material; The number of the speech that C2 branches away for the participle device; C3 is the number of the speech that correctly branches away of participle device, the language material of evaluation and test comes from Taobao and the store, Jingdone district the comment about cosmetics, these comment language materials obtain by reptile device and receiving trap, extract 251 comments then at random out, through artificial participle and correction, form the evaluating standard language material.Evaluation result is as shown in the table:

As can be seen from the above table, the F value of participle of the present invention has reached 94.6%, has obtained higher performance, has laid a solid foundation for generating high-quality tlv triple.

The performance of the present invention aspect feature extraction is also very excellent, the performance index of feature extraction are calculated with coverage rate (coverage), computing formula is: coverage=Four/Fall, Four is the feature sum that the present invention identifies when feature extraction, and Fall is by the feature sum behind the artificial mark.Our experiment language material comes from Taobao and the store, Jingdone district in the comment about cosmetics, and these review information are obtained by reptile device and receiving trap, extracts 1745 comments then at random out as the evaluation and test language material.These language materials add up to 398 through the feature after manually marking, and the feature that the present invention extracts adds up to 338, and coverage rate has reached 84.9%, has shown higher feature coverage rate.

The generation of comment tlv triple is core of the present invention and difficult point, and the present invention uses the method judging characteristic speech of machine learning and emotion speech whether can form rational collocation.In order to evaluate and test the effect that the comment tlv triple generates, from the comment of cosmetic field, randomly draw out 133 comments as test set.These test sets form the evaluating standard language material through artificial extraction comment tlv triple with after proofreading and correct.The standard of evaluation and test adopts the mode of P-R-F.Wherein C1 is the number of actual comment tlv triple in the language material; C2 is the comment tlv triple number that sorter identifies; C3 is the number of the tlv triple that correctly identifies of sorter.Evaluation result is as shown in the table:

As can be seen from the above table, the rate of accuracy reached of comment tlv triple identification is to 80.4%, and recall rate has reached 67.5%, and this has been pretty good result for the comment language material of irregularity.

The present invention also provides a kind of generation system of making a summary based on the user comment of tlv triple, this system comprises: pretreatment unit, be used to set up the feature dictionary of object, mapping vocabulary and emotion dictionary, and according to the characteristics tree of the feature construction object in the feature dictionary, wherein, mapping speech in the mapping vocabulary is corresponding with the Feature Mapping in the feature dictionary, the emotion dictionary comprises positive emotion dictionary and negative emotion dictionary, the root node on characteristics tree top is an object, each layer leaf node under the root node is the feature of object, and following one deck leaf node is the subcharacter of the last layer leaf node of correspondence, and the last layer leaf node is father's feature of following one deck leaf node of correspondence; The reptile device is used for the directed user comment webpage that grasps object from the internet; Receiving trap is used for receiving all user comments of user comment webpage object; Treating apparatus is used for each user comment is handled one by one, generates comment summary separately, and this treating apparatus comprises: the feature extraction device is used for the feature according to feature dictionary and mapping vocabulary extraction object; Emotion speech recognition device is used for according to emotion dictionary identification emotion speech; Comment tlv triple generating apparatus, be used to arrange in pairs or groups feature and emotion speech generate the comment tlv triple based on object, feature, emotion speech; Comment summary generating apparatus is used to extract the comment summary that the comment tlv triple generates this user comment; And, this system also comprises: decision-making tlv triple generating apparatus, be used to conclude the comment tlv triple of integrating all user comments, the feature that will be the comment tlv triple of feature with the leaf node below the characteristics tree ground floor leaf node sums up in the point that on its corresponding ground floor leaf node that generating all is the decision-making tlv triple of feature with characteristics tree ground floor leaf node; Calculation element, the quantity that is used for the calculated characteristics decision-making tlv triple identical with emotion speech polarity, if quantity equals 1, with this decision-making tlv triple and incompatible this decision-making tlv triple of representing of sets of numbers, if quantity is greater than 1, with any one decision-making tlv triple wherein with sets of numbers is incompatible represents that these have the decision-making tlv triple of same characteristic features and emotion speech polarity; Decision-making summary generating apparatus is used for extracting the decision-making summary that all decision-making tlv triple generate all user comments with the representation of decision-making tlv triple and number combinations.

Described feature extraction device comprises: the device that user comment is divided into sentence; Each speech in the traversal sentence, judge whether it occurs in the feature dictionary, if, extract as feature, if do not appear in the feature dictionary but appear in the mapping vocabulary, extract in the feature dictionary to come out as the device of feature with this speech mapping characteristic of correspondence.

Described emotion speech recognition device comprises: the device that user comment is divided into sentence; Traversal each speech in the sentence extracts device as the emotion speech with appearing at speech in the emotion dictionary; Judge the device of the polarity of the emotion speech that extracts according to the polarity of emotion dictionary.

Described comment tlv triple generating apparatus comprises: the device that extracts the feature templates of tlv triple from training sample; Use the device of svm classifier method according to a sorter of feature templates training; Utilize syntax rule that feature and emotion speech are made up, generate the device of tlv triple; Utilize sorter that feature and emotion speech are arranged in pairs or groups, generate the device of tlv triple; Utilize candidate's tlv triple set pair of artificial mark to filter, remove feature and emotion speech irrational tlv triple of arranging in pairs or groups, obtain commenting on the device of tlv triple by all tlv triple that syntax rule and sorter generate.

And, described calculation element also is used to calculate the quantity of positive decision-making tlv triple and the quantity of negative decision-making tlv triple, and described decision-making summary generating apparatus also is used to extract the content of the quantity of the quantity of this front decision-making tlv triple and negative decision-making tlv triple as described decision-making summary.

The correlation technique that native system adopted is identical with the embodiment of the generation method of above-mentioned user comment summary based on tlv triple, no longer repeats at this.

Design focal point of the present invention is: utilize this method or system, with the feature in each user comment, the emotion speech extracts generation based on object, feature, the comment tlv triple of emotion speech, for each user comment generates its comment summary based on the comment tlv triple, realized the information that the user paid close attention in the user comment is extracted succinctly summary info intuitively of formation separately, so that the user checks reference, and, conclude integration by commenting on tlv triple, generation has the decision-making tlv triple of directive significance, and extract all the decision-making tlv triple with the representation of decision-making tlv triple and number combinations and generate and to reflect the overall assessment result, decision-making summary with decision-making booster action, in the decision-making summary, the user can check directly that the front of own feature of being paid close attention to and further feature is estimated and how many negative evaluations respectively accounts for, also can know the overall assessment result of all user comments, thereby assisted user is made correct decisions apace.

The above, it only is preferred embodiment of the present invention, be not that technical scope of the present invention is imposed any restrictions, so every foundation technical spirit of the present invention all still belongs in the scope of technical solution of the present invention any trickle modification, equivalent variations and modification that above embodiment did.

Claims (10)

1. the generation method based on the user comment summary of tlv triple is characterized in that, comprising:
Step 1. is set up feature dictionary, mapping vocabulary and the emotion dictionary of object, and according to the characteristics tree of the feature construction object in the feature dictionary, wherein, mapping speech in the mapping vocabulary is corresponding with the Feature Mapping in the feature dictionary, the emotion dictionary comprises positive emotion dictionary and negative emotion dictionary, the root node on characteristics tree top is an object, each layer leaf node under the root node is the feature of object, and following one deck leaf node is the subcharacter of the last layer leaf node of correspondence, and the last layer leaf node is father's feature of following one deck leaf node of correspondence;
Step 2. is the directed user comment webpage that grasps object from the internet;
Step 3. receives all user comments of object in the user comment webpage;
Step 4. is carried out following processing one by one to each user comment, generates comment summary separately:
Step 41. extracts the feature of object according to feature dictionary and mapping vocabulary;
Step 42. is according to emotion dictionary identification emotion speech;
Step 43. collocation feature and emotion speech generate the comment tlv triple based on object, feature, emotion speech;
Step 44. extracts the comment summary that the comment tlv triple generates this user comment;
And this method also comprises:
Step 5. is concluded the comment tlv triple of integrating all user comments, the feature that will be the comment tlv triple of feature with the leaf node below the characteristics tree ground floor leaf node sums up in the point that on its corresponding ground floor leaf node that generating all is the decision-making tlv triple of feature with characteristics tree ground floor leaf node;
The quantity of the decision-making tlv triple that step 6. calculated characteristics is identical with emotion speech polarity, if quantity equals 1, with this decision-making tlv triple and incompatible this decision-making tlv triple of representing of sets of numbers, if quantity is greater than 1, with any one decision-making tlv triple wherein with sets of numbers is incompatible represents that these have the decision-making tlv triple of same characteristic features and emotion speech polarity;
Step 7. extracts the decision-making summary that all decision-making tlv triple generate all user comments with the representation of decision-making tlv triple and number combinations.
2. a kind of generation method of making a summary based on the user comment of tlv triple according to claim 1 is characterized in that described step 41 comprises:
Step 411. is divided into sentence with user comment;
Each speech in the step 412. traversal sentence, judge whether it occurs in the feature dictionary, if, extract as feature, if do not appear in the feature dictionary but appear in the mapping vocabulary, extract in the feature dictionary to come out as feature with this speech mapping characteristic of correspondence.
3. a kind of generation method of making a summary based on the user comment of tlv triple according to claim 1 is characterized in that described step 42 comprises:
Step 421. is divided into sentence with user comment;
Each speech in the step 422. traversal sentence extracts the speech that appears in the emotion dictionary as the emotion speech;
Step 423. is judged the polarity of the emotion speech that extracts according to the polarity of emotion dictionary.
4. a kind of generation method of making a summary based on the user comment of tlv triple according to claim 1 is characterized in that described step 43 comprises:
Step 431. is extracted the feature templates of tlv triple from training sample;
Step 432. uses the svm classifier method according to sorter of feature templates training;
Step 433. utilizes syntax rule that feature and emotion speech are made up, and generates tlv triple;
Step 434. utilizes sorter that feature and emotion speech are arranged in pairs or groups, and generates tlv triple;
Step 435. utilizes candidate's tlv triple set pair of artificial mark to be filtered by all tlv triple that syntax rule and sorter generate, and removes feature and emotion speech irrational tlv triple of arranging in pairs or groups, and obtains commenting on tlv triple.
5. a kind of generation method of making a summary according to claim 1 based on the user comment of tlv triple, it is characterized in that, described step 6 also comprises the quantity of calculating positive decision-making tlv triple and the quantity of negative decision-making tlv triple, and described step 7 also comprises the content of the quantity of the quantity that extracts this front decision-making tlv triple and negative decision-making tlv triple as described decision-making summary.
6. the generation system based on the user comment summary of tlv triple is characterized in that, comprising:
Pretreatment unit, be used to set up feature dictionary, mapping vocabulary and the emotion dictionary of object, and according to the characteristics tree of the feature construction object in the feature dictionary, wherein, mapping speech in the mapping vocabulary is corresponding with the Feature Mapping in the feature dictionary, the emotion dictionary comprises positive emotion dictionary and negative emotion dictionary, the root node on characteristics tree top is an object, each layer leaf node under the root node is the feature of object, and following one deck leaf node is the subcharacter of the last layer leaf node of correspondence, and the last layer leaf node is father's feature of following one deck leaf node of correspondence;
The reptile device is used for the directed user comment webpage that grasps object from the internet;
Receiving trap is used for receiving all user comments of user comment webpage object;
Treating apparatus is used for each user comment is handled one by one, generates comment summary separately, and this treating apparatus comprises:
The feature extraction device is used for the feature according to feature dictionary and mapping vocabulary extraction object;
Emotion speech recognition device is used for according to emotion dictionary identification emotion speech;
Comment tlv triple generating apparatus, be used to arrange in pairs or groups feature and emotion speech generate the comment tlv triple based on object, feature, emotion speech;
Comment summary generating apparatus is used to extract the comment summary that the comment tlv triple generates this user comment;
And this system also comprises:
Decision-making tlv triple generating apparatus, be used to conclude the comment tlv triple of integrating all user comments, the feature that will be the comment tlv triple of feature with the leaf node below the characteristics tree ground floor leaf node sums up in the point that on its corresponding ground floor leaf node that generating all is the decision-making tlv triple of feature with characteristics tree ground floor leaf node;
Calculation element, the quantity that is used for the calculated characteristics decision-making tlv triple identical with emotion speech polarity, if quantity equals 1, with this decision-making tlv triple and incompatible this decision-making tlv triple of representing of sets of numbers, if quantity is greater than 1, with any one decision-making tlv triple wherein with sets of numbers is incompatible represents that these have the decision-making tlv triple of same characteristic features and emotion speech polarity;
Decision-making summary generating apparatus is used for extracting the decision-making summary that all decision-making tlv triple generate all user comments with the representation of decision-making tlv triple and number combinations.
7. a kind of generation system of making a summary based on the user comment of tlv triple according to claim 6 is characterized in that described feature extraction device comprises:
The device that user comment is divided into sentence;
Each speech in the traversal sentence, judge whether it occurs in the feature dictionary, if, extract as feature, if do not appear in the feature dictionary but appear in the mapping vocabulary, extract in the feature dictionary to come out as the device of feature with this speech mapping characteristic of correspondence.
8. a kind of generation system of making a summary based on the user comment of tlv triple according to claim 6 is characterized in that described emotion speech recognition device comprises:
The device that user comment is divided into sentence;
Traversal each speech in the sentence extracts device as the emotion speech with appearing at speech in the emotion dictionary;
Judge the device of the polarity of the emotion speech that extracts according to the polarity of emotion dictionary.
9. a kind of generation system of making a summary based on the user comment of tlv triple according to claim 6 is characterized in that described comment tlv triple generating apparatus comprises:
From training sample, extract the device of the feature templates of tlv triple;
Use the device of svm classifier method according to a sorter of feature templates training;
Utilize syntax rule that feature and emotion speech are made up, generate the device of tlv triple;
Utilize sorter that feature and emotion speech are arranged in pairs or groups, generate the device of tlv triple;
Utilize candidate's tlv triple set pair of artificial mark to filter, remove feature and emotion speech irrational tlv triple of arranging in pairs or groups, obtain commenting on the device of tlv triple by all tlv triple that syntax rule and sorter generate.
10. a kind of generation system of making a summary according to claim 6 based on the user comment of tlv triple, it is characterized in that, described calculation element also is used to calculate the quantity of positive decision-making tlv triple and the quantity of negative decision-making tlv triple, and described decision-making summary generating apparatus also is used to extract the content of the quantity of the quantity of this front decision-making tlv triple and negative decision-making tlv triple as described decision-making summary.
CN201110236683.7A 2011-08-17 2011-08-17 Method and system for generating user comment summaries based on triples CN102236722B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201110236683.7A CN102236722B (en) 2011-08-17 2011-08-17 Method and system for generating user comment summaries based on triples

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201110236683.7A CN102236722B (en) 2011-08-17 2011-08-17 Method and system for generating user comment summaries based on triples

Publications (2)

Publication Number Publication Date
CN102236722A true CN102236722A (en) 2011-11-09
CN102236722B CN102236722B (en) 2014-08-27

Family

ID=44887368

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201110236683.7A CN102236722B (en) 2011-08-17 2011-08-17 Method and system for generating user comment summaries based on triples

Country Status (1)

Country Link
CN (1) CN102236722B (en)

Cited By (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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
CN102945268A (en) * 2012-10-25 2013-02-27 北京腾逸科技发展有限公司 Method and system for excavating comments on characteristics of product
CN103377262A (en) * 2012-04-28 2013-10-30 国际商业机器公司 Method and device for grouping users
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
CN103678371A (en) * 2012-09-14 2014-03-26 富士通株式会社 Lexicon updating device, data integration device and method and electronic device
CN103970783A (en) * 2013-01-31 2014-08-06 百度在线网络技术(北京)有限公司 LBS (Location Based Service)-based information acquisition method and equipment
CN103970784A (en) * 2013-01-31 2014-08-06 百度在线网络技术(北京)有限公司 Retrieval method and equipment
CN103970786A (en) * 2013-01-31 2014-08-06 百度在线网络技术(北京)有限公司 LBS (Location Based Service)-based information obtaining method and equipment
CN104375739A (en) * 2013-08-12 2015-02-25 联想(北京)有限公司 Information processing method and electronic equipment
CN104375977A (en) * 2013-08-14 2015-02-25 腾讯科技(深圳)有限公司 Answer message processing method and device for question-answer communities
CN104462132A (en) * 2013-09-23 2015-03-25 华为技术有限公司 Comment information display method and device
CN105512333A (en) * 2015-12-28 2016-04-20 上海电机学院 Product comment theme searching method based on emotional tendency
CN105760502A (en) * 2016-02-23 2016-07-13 常州普适信息科技有限公司 Commercial quality emotional dictionary construction system based on big data text mining
CN105761152A (en) * 2016-02-07 2016-07-13 重庆邮电大学 Topic participation prediction method based on triadic group in social network
CN105912644A (en) * 2016-04-08 2016-08-31 国家计算机网络与信息安全管理中心 Network review generation type abstract method
CN106055542A (en) * 2016-08-17 2016-10-26 山东大学 Automatic text summarization generation method and automatic text summarization generation system based on temporal knowledge extraction
CN106469145A (en) * 2016-09-30 2017-03-01 中科鼎富(北京)科技发展有限公司 Text emotion analysis method and device
CN110349620A (en) * 2019-06-28 2019-10-18 广州序科码生物技术有限责任公司 One kind accurately identifying interaction of molecules and its polarity and directionality method from PubMed document

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101667194A (en) * 2009-09-29 2010-03-10 北京大学 Automatic abstracting method and system based on user comment text feature
CN101727487A (en) * 2009-12-04 2010-06-09 中国人民解放军信息工程大学 Network criticism oriented viewpoint subject identifying method and system
CN102096680A (en) * 2009-12-15 2011-06-15 北京大学 Method and device for analyzing information validity

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101667194A (en) * 2009-09-29 2010-03-10 北京大学 Automatic abstracting method and system based on user comment text feature
CN101727487A (en) * 2009-12-04 2010-06-09 中国人民解放军信息工程大学 Network criticism oriented viewpoint subject identifying method and system
CN102096680A (en) * 2009-12-15 2011-06-15 北京大学 Method and device for analyzing information validity

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
郗亚辉等: "产品评论挖掘研究综述", 《山东大学学报(理学版)》 *

Cited By (22)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103377262A (en) * 2012-04-28 2013-10-30 国际商业机器公司 Method and device for grouping users
CN103377262B (en) * 2012-04-28 2017-09-12 国际商业机器公司 The method and apparatus being grouped to user
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
CN103678371A (en) * 2012-09-14 2014-03-26 富士通株式会社 Lexicon updating device, data integration device and method and electronic device
CN103678371B (en) * 2012-09-14 2017-10-10 富士通株式会社 Word library updating device, data integration device and method and electronic equipment
CN102945268A (en) * 2012-10-25 2013-02-27 北京腾逸科技发展有限公司 Method and system for excavating comments on characteristics of product
CN103970784A (en) * 2013-01-31 2014-08-06 百度在线网络技术(北京)有限公司 Retrieval method and equipment
CN103970786A (en) * 2013-01-31 2014-08-06 百度在线网络技术(北京)有限公司 LBS (Location Based Service)-based information obtaining method and equipment
CN103970783A (en) * 2013-01-31 2014-08-06 百度在线网络技术(北京)有限公司 LBS (Location Based Service)-based information acquisition method and equipment
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
CN104375739A (en) * 2013-08-12 2015-02-25 联想(北京)有限公司 Information processing method and electronic equipment
CN104375977A (en) * 2013-08-14 2015-02-25 腾讯科技(深圳)有限公司 Answer message processing method and device for question-answer communities
CN104375977B (en) * 2013-08-14 2018-11-23 腾讯科技(深圳)有限公司 The processing method and processing device of reply message in Ask-Answer Community
CN104462132A (en) * 2013-09-23 2015-03-25 华为技术有限公司 Comment information display method and device
CN105512333A (en) * 2015-12-28 2016-04-20 上海电机学院 Product comment theme searching method based on emotional tendency
CN105761152A (en) * 2016-02-07 2016-07-13 重庆邮电大学 Topic participation prediction method based on triadic group in social network
CN105760502A (en) * 2016-02-23 2016-07-13 常州普适信息科技有限公司 Commercial quality emotional dictionary construction system based on big data text mining
CN105912644A (en) * 2016-04-08 2016-08-31 国家计算机网络与信息安全管理中心 Network review generation type abstract method
CN106055542A (en) * 2016-08-17 2016-10-26 山东大学 Automatic text summarization generation method and automatic text summarization generation system based on temporal knowledge extraction
CN106055542B (en) * 2016-08-17 2019-01-22 山东大学 A kind of text snippet automatic generation method and system based on temporal knowledge extraction
CN106469145A (en) * 2016-09-30 2017-03-01 中科鼎富(北京)科技发展有限公司 Text emotion analysis method and device
CN110349620A (en) * 2019-06-28 2019-10-18 广州序科码生物技术有限责任公司 One kind accurately identifying interaction of molecules and its polarity and directionality method from PubMed document

Also Published As

Publication number Publication date
CN102236722B (en) 2014-08-27

Similar Documents

Publication Publication Date Title
CN103778214B (en) A kind of item property clustering method based on user comment
EP3096246A1 (en) Method, system and storage medium for realizing intelligent answering of questions
CN104317959B (en) Data digging method based on social platform and device
Stamatatos et al. Overview of the pan/clef 2015 evaluation lab
CN103593054B (en) A kind of combination Emotion identification and the question answering system of output
US20140195348A1 (en) Method and apparatus for composing search phrases, distributing ads and searching product information
CN102929937B (en) Based on the data processing method of the commodity classification of text subject model
CN104049755B (en) Information processing method and device
JP5827416B2 (en) User question processing method and processing system
CN105279495A (en) Video description method based on deep learning and text summarization
CN103714139A (en) Parallel data mining method for identifying a mass of mobile client bases
JP2013531847A (en) Intelligent navigation method, apparatus and system
CN103778260A (en) Individualized microblog information recommending system and method
CN103917968A (en) System and method for managing opinion networks with interactive opinion flows
CN104408093B (en) A kind of media event key element abstracting method and device
CN104636425B (en) A kind of network individual or colony's Emotion recognition ability prediction and method for visualizing
CN105930503A (en) Combination feature vector and deep learning based sentiment classification method and device
CN102156737B (en) Method for extracting subject content of Chinese webpage
Salloum et al. Mining social media text: extracting knowledge from Facebook
CN103413550A (en) Man-machine interactive language learning system and method
CN103500175B (en) A kind of method based on sentiment analysis on-line checking microblog hot event
CN103336766B (en) Short text garbage identification and modeling method and device
CN102314489B (en) Method for analyzing opinion leader in network forum
CN102929928A (en) Multidimensional-similarity-based personalized news recommendation method
Celli Unsupervised personality recognition for social network sites

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant