CN109800418A - Text handling method, device and storage medium - Google Patents
Text handling method, device and storage medium Download PDFInfo
- Publication number
- CN109800418A CN109800418A CN201811539984.5A CN201811539984A CN109800418A CN 109800418 A CN109800418 A CN 109800418A CN 201811539984 A CN201811539984 A CN 201811539984A CN 109800418 A CN109800418 A CN 109800418A
- Authority
- CN
- China
- Prior art keywords
- emotion
- collocation
- collocation group
- word
- candidate
- 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.)
- Granted
Links
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Machine Translation (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention provides a kind of text handling method, device and storage medium, this method comprises: according to the existing user comment text of target domain, obtain the emotion dictionary of collocations of target domain, it include multiple target emotion collocation groups in emotion dictionary of collocations, each target emotion collocation group is used to characterize user and comments on the emotion of the attribute of the comment object of target domain;According to the user comment text to be processed and emotion dictionary of collocations of target domain, the corresponding emotion collocation group of user comment text to be processed is obtained.Text handling method provided by the invention constructs the emotion dictionary of collocations of target domain according to existing user comment text, then the emotion collocation group of text to be processed is obtained using the emotion dictionary of collocations, can accurately obtain the emotion viewpoint of the text of target domain.
Description
Technical field
The present invention relates to text emotion analysis technical fields more particularly to a kind of text handling method, device and storage to be situated between
Matter.
Background technique
User comment text (UGC text) is that user experience crosses the text evaluated after product product, in text
Emotion viewpoint extract it is most important;If automatic comment text is the comment text to a dining room, including " the taste in this family restaurant
Road is pretty good ", " I likes well taste here " and " steamed beef soup of this family is drunk very well ", to the emotion viewpoint of above-mentioned comment text
Extracting result is " taste, good, actively ";To in text emotion viewpoint extract result can make businessman see user for
The impression of oneself product, more targeted carry out product optimization, and also allow user by comparing commenting for different businessmans
By preferably carrying out consumption decision.
In the prior art, the emotion viewpoint that can be realized field of general technology extracts, but due in field of general technology
For the attribute word of emotion viewpoint for pervasive every field, the attribute word of emotion viewpoint therein is excessively single, is not particularly suited for hanging down
Straight field;If the extraction result of the emotion viewpoint of field of general technology is " design, good, actively ", and not applicable cuisines field
Comment text " taste in this family restaurant is pretty good ".
Summary of the invention
The present invention provides a kind of text handling method, device and storage medium, is constructed according to existing user comment text
The emotion dictionary of collocations of target domain, then the emotion collocation group for using the emotion dictionary of collocations to obtain text to be processed, Neng Gouzhun
Really obtain the emotion viewpoint of the text of target domain.
A kind of offer text handling method of the first aspect of the present invention, comprising:
According to the existing user comment text of target domain, the emotion dictionary of collocations of the target domain is obtained, it is described
It include multiple target emotion collocation groups in emotion dictionary of collocations, each target emotion collocation group is for characterizing user to the target
The emotion comment of the attribute of the comment object in field;
According to the user comment text to be processed of the target domain and the emotion dictionary of collocations, obtain described to be processed
The corresponding emotion collocation group of user comment text.
Optionally, the target emotion collocation group includes dimension word and evaluating word, and the dimension word is the existing use
The attribute of comment object in the comment text of family;The existing user comment text according to target domain, obtains the mesh
The emotion dictionary of collocations in mark field, comprising:
Word segmentation processing is carried out to each existing user comment text, obtains each existing user comment text
This multiple words;
According to the corresponding part of speech of multiple words of each existing user comment text, and, part of speech collocation rule,
The first candidate emotion collocation group of each existing user comment text is obtained, the part of speech collocation rule includes: dimension
Word is established rules really, and evaluating word is established rules then really;
According to the multiple described first candidate emotion collocation groups, the emotion dictionary of collocations of the target domain is obtained.
Optionally, the target emotion collocation group further includes emotion word, and the emotion word is the existing user comment
The feeling polarities of text;After the first candidate emotion collocation group for obtaining each existing user comment text, also
Include:
Feeling polarities analysis is carried out to each described first candidate emotion collocation group, obtains each described first candidate emotion
The corresponding emotion word of collocation group;
It is described according to multiple first candidate emotion collocation groups, obtain the emotion dictionary of collocations of the target domain, wrap
It includes:
According to each described first candidate emotion collocation group and the corresponding emotion of each first candidate emotion collocation group
Word obtains the emotion dictionary of collocations of the target domain.
Optionally, described according to each described first candidate emotion collocation group and each described first candidate emotion collocation group
Corresponding emotion word obtains the emotion dictionary of collocations of the target domain, comprising:
By each described first candidate emotion collocation group and the corresponding emotion word of each first candidate emotion collocation group
It is combined, obtains the second candidate emotion collocation group, each described second candidate emotion collocation group includes the dimension word, described
Evaluating word and the emotion word;
It is semantic according to the first semantic and evaluating word second of the dimension word of each described second candidate emotion collocation group, it is right
It is clustered with identical first semantic and the second semanteme the second candidate emotion collocation group, obtains the collocation of third candidate emotion
Group;
According to the quantity of the corresponding second candidate emotion collocation group of each third candidate emotion collocation group, and according to institute
It states the sequence of quantity from big to small to be ranked up, third candidate's emotion collocation group of preset quantity will be arranged in front as described in
The emotion dictionary of collocations of target domain.
Optionally, described that feeling polarities analysis is carried out to each described first candidate emotion collocation group, it obtains each described
Before the corresponding emotion word of first candidate's emotion collocation group, further includes:
Multiple described first candidate emotion collocation groups are screened, deletes and does not meet the first candidate for presetting interdependent rule
Emotion collocation group, it is described to preset interdependent rule are as follows: evaluating word and dimension word in the described first candidate emotion collocation group exist dynamic
Guest's relationship, and/or, the dimension word and the described first candidate emotion collocation group in the described first candidate emotion collocation group are corresponding
With the presence of user comment text in comment object subject-predicate relationship.
Optionally, the user comment text to be processed and the emotion dictionary of collocations according to the target domain, is obtained
Take the corresponding emotion collocation group of the user comment text to be processed, comprising:
If including in first object emotion collocation group in the emotion dictionary of collocations in the user comment text to be processed
Dimension word and evaluating word, then using the first object emotion collocation group as the corresponding feelings of the user comment text to be processed
Feel collocation group, the first object emotion collocation group is any one target emotion collocation group in the emotion dictionary of collocations;
If in the user comment text to be processed only including the second target emotion collocation group in the emotion dictionary of collocations
In dimension word, and the emotion pole of the feeling polarities of the user comment text to be processed and the second target emotion collocation group
Property is identical, then using the second target emotion collocation group as the corresponding emotion collocation group of the user comment text to be processed,
The second target emotion collocation group is any one target emotion collocation group in the emotion dictionary of collocations.
Optionally, the user comment text to be processed and the emotion dictionary of collocations according to the target domain, is obtained
Take the corresponding emotion collocation group of the user comment text to be processed, comprising:
If do not include in the user comment text to be processed in the emotion dictionary of collocations any one target emotion take
Dimension word in combo will then be greater than the target feelings of similarity threshold with the semantic similarity of the user comment text to be processed
Collocation group is felt as the corresponding emotion collocation group of the user comment text to be processed.
The second aspect of the present invention provides a kind of text processing apparatus, comprising:
Emotion dictionary of collocations obtains module and obtains the mesh for the existing user comment text according to target domain
The emotion dictionary of collocations in mark field, includes multiple target emotion collocation groups in the emotion dictionary of collocations, and each target emotion is taken
Combo is used to characterize user and comments on the emotion of the attribute of the comment object of the target domain;
Emotion collocation group obtain module, for according to the target domain user comment text to be processed and the emotion
Dictionary of collocations obtains the corresponding emotion collocation group of the user comment text to be processed.
Optionally, the target emotion collocation group includes dimension word and evaluating word, and the dimension word is the existing use
The attribute of comment object in the comment text of family.
Optionally, the emotion dictionary of collocations obtains module, is specifically used for each existing user comment text
Word segmentation processing is carried out, multiple words of each existing user comment text are obtained;According to each existing user
The corresponding part of speech of multiple words of comment text, and, part of speech collocation rule obtains each existing user comment text
The first candidate emotion collocation group, the part of speech collocation rule includes: that dimension word is established rules really, and evaluating word is established rules then really;
According to the multiple described first candidate emotion collocation groups, the emotion dictionary of collocations of the target domain is obtained.
Optionally, the target emotion collocation group further includes emotion word, and the emotion word is the existing user comment
The feeling polarities of text.
Optionally, described device further include: emotion word obtains module;
The emotion word obtains module, for carrying out feeling polarities analysis to each described first candidate emotion collocation group,
Obtain the corresponding emotion word of each first candidate emotion collocation group.
Optionally, emotion dictionary of collocations obtains module, be specifically used for according to each first candidate emotion collocation group and
The corresponding emotion word of each first candidate emotion collocation group, obtains the emotion dictionary of collocations of the target domain.
Optionally, emotion dictionary of collocations obtains module, specifically for by each described first candidate emotion collocation group and often
The corresponding emotion word of a first candidate emotion collocation group is combined, and obtains the second candidate emotion collocation group, each described
Second candidate emotion collocation group includes the dimension word, the evaluating word and the emotion word;It is candidate according to each described second
Semantic the second semanteme with evaluating word of the first of the dimension word of emotion collocation group, to semantic and the second semanteme with identical first
Second candidate emotion collocation group is clustered, and third candidate emotion collocation group is obtained;It is taken according to each third candidate emotion
The quantity of the corresponding second candidate emotion collocation group of combo, and be ranked up according to the sequence of the quantity from big to small, it will arrange
Emotion dictionary of collocations of the third candidate emotion collocation group of preset quantity as the target domain before being listed in.
Optionally, described device further include: removing module;
The removing module, for screening to the multiple described first candidate emotion collocation groups, deletion does not meet default
The candidate emotion collocation group of the first of interdependent rule, it is described to preset interdependent rule are as follows: commenting in the described first candidate emotion collocation group
There is dynamic guest's relationship in valence word and dimension word, and/or, the dimension word and described first in the described first candidate emotion collocation group are candidate
There are subject-predicate relationships for comment object in the corresponding existing user comment text of emotion collocation group.
Optionally, the emotion collocation group obtains module, if specifically for including in the user comment text to be processed
Dimension word and evaluating word in the emotion dictionary of collocations in first object emotion collocation group, then take the first object emotion
For combo as the corresponding emotion collocation group of the user comment text to be processed, the first object emotion collocation group is the feelings
Feel any one target emotion collocation group in dictionary of collocations;If in the user comment text to be processed only including the emotion
Dimension word in dictionary of collocations in the second target emotion collocation group, and the feeling polarities of the user comment text to be processed and institute
The feeling polarities for stating the second target emotion collocation group are identical, then using the second target emotion collocation group as the use to be processed
The corresponding emotion collocation group of family comment text, the second target emotion collocation group are any one in the emotion dictionary of collocations
A target emotion collocation group.
Optionally, the emotion collocation group obtains module, if specifically for not wrapping in the user comment text to be processed
Include the dimension word in the emotion dictionary of collocations in any one target emotion collocation group, then it will be with the user comment to be processed
The target emotion collocation group that the semantic similarity of text is greater than similarity threshold is corresponding as the user comment text to be processed
Emotion collocation group.
The third aspect of the present invention provides a kind of text processing apparatus, comprising: at least one processor and memory;
The memory stores computer executed instructions;
At least one described processor executes the computer executed instructions of the memory storage, so that the text-processing
Device executes above-mentioned text handling method.
The fourth aspect of the present invention provides a kind of computer readable storage medium, deposits on the computer readable storage medium
Computer executed instructions are contained, when the computer executed instructions are executed by processor, realize above-mentioned text handling method.
The present invention provides a kind of text handling method, device and storage medium, this method comprises: according to target domain
Some user comment texts, obtain the emotion dictionary of collocations of target domain, include that multiple target emotions are taken in emotion dictionary of collocations
Combo, each target emotion collocation group are used to characterize user and comment on the emotion of the attribute of the comment object of target domain;According to
The user comment text to be processed and emotion dictionary of collocations of target domain obtain the corresponding emotion of user comment text to be processed and take
Combo.Text handling method provided by the invention constructs the emotion Collocation of target domain according to existing user comment text
Allusion quotation, then the emotion collocation group for using the emotion dictionary of collocations to obtain text to be processed, can accurately obtain the text of target domain
Emotion viewpoint.
Detailed description of the invention
Fig. 1 is the flow diagram one of text handling method provided by the invention;
Fig. 2 is the Comparative result exemplary diagram of the prior art and text handling method provided by the invention;
Fig. 3 is the flow diagram two of text handling method provided by the invention;
Fig. 4 is the structural schematic diagram one of text processing apparatus provided by the invention;
Fig. 5 is the structural schematic diagram two of text processing apparatus provided by the invention;
Fig. 6 is the structural schematic diagram three of text processing apparatus provided by the invention.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with the embodiment of the present invention, to this
Technical solution in inventive embodiments is clearly and completely described, it is clear that described embodiment is that a part of the invention is real
Example is applied, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not making creation
Property labour under the premise of every other embodiment obtained, shall fall within the protection scope of the present invention.
The excavation of emotion comment viewpoint is intended to extract the emotion viewpoint information in user comment.For given product (packet
Include commodity and service) user comment text, automatically analyze text perfusion dimension (service, room, the traffic in such as hotel) and
It comments on viewpoint (such as good, simple and crude, convenient), output comment viewpoint label (service is pretty good, room is simple and crude, traffic convenience), and
Comment on the feeling polarities of viewpoint (favorable comment or difference are commented).Businessman can be helped to carry out product point based on emotion comment opining mining
Analysis, auxiliary user carry out consumption decision.
A kind of text of the user comment text (User Generated Content, UGC) as special shape, emotion
Showing also for viewpoint has many particularity.Firstly, dimension is varied under different vertical field, the i.e. vertical field of dimension
Comment on attribute;The dimension that user is concerned about in such as " cuisines " vertical class essentially consists in " taste ", " environment ", " service ";And in " religion
Educate " hang down class under user more concern that dimension is " quality of education ", " cost performance " etc..Secondly, for same dimension word
Description, the expression way of emotion viewpoint is diversified.Emotion viewpoint as expressed " taste is pretty good " in " cuisines " vertical class,
The description of its text can be that " taste in this family dining room is pretty good ", " I likes well taste here ", " steamed beef soup of this family is fine
Drink " etc..
The extraction of emotion viewpoint in the prior art to user comment text is often embodied in field of general technology, and by
Can not effectively identify the dimension word in specific area in the extraction of the emotion viewpoint of field of general technology, thus cause recall rate without
Method reaches requirement.Fig. 2 is the Comparative result exemplary diagram of the prior art and text handling method provided by the invention, as shown in Fig. 2,
The user comment text that " automobile " hangs down under class be " this vehicle combination automobile partner position appearance and structure and design, be it is a independently
Brand and the high vehicle of vehicle safety, the fuel economy of this vehicle is very outstanding, and operating is also more smooth and quiet.But come relatively
It says that the cross-country ability of this vehicle is most strong, the SUV open car formula riding experience of great individual character is provided, the repacking of vehicle is also quite big
", use the extraction result of the emotion viewpoint of field of general technology in the prior art for " design, good " and " security performance,
It is high ".It can not extract the dimension that user is concerned about under " automobile " vertical class, such as " cross-country ability ", " fuel economy " and " driving body
Test " etc..
In order to which the more accurate emotion viewpoint in user comment text extracts, the present invention provides a kind of texts
Processing method, Fig. 1 are the flow diagram one of text handling method provided by the invention, the execution master of method flow shown in Fig. 1
Body can be text processing apparatus, and text processing unit can be by arbitrary software and or hardware realization.As shown in Figure 1, this reality
Applying the text handling method that example provides may include:
S101 obtains the emotion dictionary of collocations of target domain, emotion according to the existing user comment text of target domain
It include multiple target emotion collocation groups in dictionary of collocations, each target emotion collocation group comments target domain for characterizing user
By the emotion comment of the attribute of object.
Target domain in the present embodiment can be any one field in general field, such as " cuisines " field, " vapour
Vehicle " field, " tourism " field etc.;Treatment process of the text handling method in each field in the present embodiment is identical.Wherein,
The existing user comment text of target domain can come from different databases, i.e., the text processing apparatus in the present embodiment can
It is connect with the database of the storage user comment text with target domain, the existing user in available each database comments
Paper sheet;Alternatively, the existing user comment text of target domain, which can be technical staff's collection, is directed into this article present treatment dress
It sets.
Such as: the existing user comment text in " cuisines " field can be in " XX comment " corresponding server of application program
User comment text, and/or, user comment text and/or any social activity in the corresponding server of " XX meters " application program
The comment text of user in software for " cuisines " field.
Existing user comment text can be a word, the one section of word that user edits, or by text processing apparatus
According to the one of input section of recording or the text of one section of video conversion, it is contemplated that will record or video conversion is text
Originally conversion regime in the prior art can be used;Text processing apparatus is according to existing user comment text in the present embodiment,
The emotion dictionary of collocations of target domain is obtained, includes multiple target emotion collocation groups, each target feelings in the emotion dictionary of collocations
Sense collocation group is used to characterize user and comments on the emotion of the attribute of the comment object of target domain.
Specifically, text processing apparatus converts user comment text to the mode of target emotion collocation group, target is formed
The emotion dictionary of collocations in field.Each target emotion collocation group is the emotion comment to the attribute of comment object;Illustratively, exist
In " cuisines " field, comment object can be specific a certain dining room, comment on object attribute can for " taste ", " environment ",
" service " etc.;And the target emotion collocation group that the emotion comment of the attribute of comment object is constituted can for " taste, good ",
" environment, difference ", " service, be general ".
It is envisioned that can also include the feeling polarities point to user comment text in each target emotion collocation group
As a result, as being provided with feeling polarities template in the text processing apparatus in the present embodiment, feeling polarities can be divided into " positive " for analysis
" passiveness ", it is corresponding, multiple words are respectively included in " positive " template and " passiveness " template, if the word in user comment text
Language exists in " positive " template, then sets " product for the feeling polarities of the corresponding target emotion collocation group of user comment text
Pole ";Similarly, if the word in user comment text exists in " passiveness " template, by the corresponding mesh of user comment text
The feeling polarities of mark emotion collocation group are set as " passiveness ";The each target in emotion dictionary of collocations then obtained in the present embodiment
Emotion collocation group includes feeling polarities.
Illustratively, above-mentioned " taste, good ", " environment, difference ", " service, general " corresponding target emotion collocation
Group is " taste, good, actively ", " environment, difference, passiveness ", " service, general, passive ".
S102 obtains user to be processed and comments according to the user comment text to be processed and emotion dictionary of collocations of target domain
The corresponding emotion collocation group of paper sheet.
In the present embodiment, text processing apparatus is arranged in pairs or groups after the emotion dictionary of collocations of building target domain using the emotion
The corresponding emotion collocation group of the available text to be processed of dictionary.Specifically, due to user comment text to be processed and emotion
Dictionary of collocations belongs to same target domain, user comment of the emotion dictionary of collocations obtained in the present embodiment for the target domain
Text has preferable applicability.
Wherein, it due to including multiple target emotion collocation groups in emotion dictionary of collocations, is wrapped in each target emotion collocation group
Emotion comment containing user, if in user comment text to be processed including the emotion in any one target emotion collocation group
When commenting on word, using the target emotion collocation group as the emotion collocation group of user comment text to be processed.
If not including the emotion comment word having in any one target emotion collocation group in user comment text to be processed
When, the semantic similarity of available user comment text to be processed and each target emotion collocation group, by maximum similarity pair
Emotion collocation group of the target emotion collocation group answered as user comment text to be processed.
Illustratively, if user comment text to be processed is " this family's taste is pretty good ", the target in emotion dictionary of collocations
Emotion collocation group is " taste, good, actively ", " environment, difference, passiveness ", " service, general, passive ", then can will " taste, no
It is wrong, actively " emotion collocation group as the user comment text to be processed;If user comment text to be processed is " speed of serving
Slowly, much time has been waited ", there is word in target emotion collocation group due to not including in user comment text to be processed, then basis
Semantic similarity " can will service, is general, is passive " emotion collocation group as the user comment text to be processed.
As shown in Fig. 2, use document processing method in the present embodiment carry out emotion viewpoint extraction result be " design,
Well ", " security performance, height ", " cross-country ability, strong ", " fuel economy, outstanding ", " operating, smooth ", " riding experience, no
It is wrong " and " repacking property, greatly ".
Text handling method provided in this embodiment includes: the existing user comment text according to target domain, is obtained
The emotion dictionary of collocations of target domain includes multiple target emotion collocation groups in emotion dictionary of collocations, each target emotion collocation
Group comments on the emotion of the attribute of the comment object of target domain for characterizing user;It is commented according to the user to be processed of target domain
Paper sheet and emotion dictionary of collocations obtain the corresponding emotion collocation group of user comment text to be processed.Text provided in this embodiment
Treatment method constructs the emotion dictionary of collocations of target domain according to existing user comment text, then uses the emotion Collocation
Allusion quotation obtains the emotion collocation group of text to be processed, can accurately obtain the emotion viewpoint of the text of target domain.
On the basis of the above embodiments, below with reference to Fig. 3 to how being constructed in text handling method provided by the invention
The emotion dictionary of collocations of target domain and the emotion collocation group for obtaining text to be processed are described in detail, and Fig. 3 mentions for the present invention
The flow diagram two of the text handling method of confession, as shown in figure 3, text handling method provided in this embodiment may include:
S301 carries out word segmentation processing to each existing user comment text, obtains each existing user comment text
Multiple words.
In the present embodiment, the existing user comment text of target domain be it is multiple, in order to obtain the emotion of target domain
Dictionary of collocations needs to obtain the corresponding emotion collocation group of each existing user comment text;Specifically, text processing apparatus pair
Each existing user comment text carries out word segmentation processing, wherein if existing user comment text is that long sentence or one section are talked about,
Long sentence or one section of word first can be cut into short sentence according to slit mode in the prior art;By existing user comment text
It is cut into after short sentence and carries out word segmentation processing again, obtain multiple words of each existing user comment text.
It, can also be it is worth noting that, in order to obtain the correct emotion collocation group of user comment text in the present embodiment
The clause in user comment text is filtered before carrying out word segmentation processing;Such as by the interrogative sentence and negative in user comment text
Sentence is deleted, then user comment text carries out word segmentation processing by treated.
Illustratively, existing user comment text is that " taste in this family dining room is pretty good, it is not known that can or can not someone
? ", text processing apparatus is first filtered the clause in user comment text, by question sentence " do not know can or can not someone come
" delete, will treated " taste in this family dining room is pretty good " carries out word segmentation processing, such as obtain existing user comment text
This multiple words be " this family dining room ", " ", " taste " and " pretty good ".
S302, according to the corresponding part of speech of multiple words of each existing user comment text, and, part of speech collocation rule
Then, obtain the first candidate emotion collocation group of each existing user comment text, part of speech collocation rule include: dimension word really
Set pattern is then established rules then with evaluating word really.
In the present embodiment, text processing apparatus, can be with after the multiple words for obtaining each existing user comment text
Obtain the part of speech of each word;As multiple words " this family dining room ", " ", " taste " and " pretty good " corresponding part of speech be respectively
Noun, conjunction, noun, adjective.
Target emotion collocation group in the present embodiment includes dimension word and evaluating word, wherein dimension word is existing user
The attribute of comment object in comment text.Dimension word and the evaluation in existing user comment text are obtained in the present embodiment
Word, i.e., the first candidate emotion collocation group in composition user comment text.
Specifically, part of speech collocation rule is previously stored in text processing apparatus, specifically, part of speech collocation rule packet
Include: dimension word is established rules really, and evaluating word is established rules then really;Illustratively, the part of speech collocation rule in the present embodiment can be with
Are as follows: using the noun of two words or more as dimension word, using the adjective of a word or more as evaluating word.Wherein, dimension
Word is established rules then really as the noun of two words or more, and evaluating word is established rules then really as the adjective of word or more.It can be with
It is contemplated that those skilled in the art can also be used in other part of speech collocation existing user comment texts of Rule
Dimension word and evaluating word.
Illustratively, as above-mentioned multiple words " this family dining room ", " ", " taste " and " pretty good " corresponding part of speech difference
For noun, conjunction, noun, adjective;Wherein, the noun of two words or more be " this family dining room ", " taste ", a word and with
On adjective be " pretty good ", therefore " pretty good " can will be used as evaluating word, and since " this family dining room " is that this is existing
Comment object in user comment text, therefore the existing user comment text that " taste " is used as dimension word, therefore is obtained
The candidate emotion collocation group of this first is " taste, pretty good ".
S303 screens the multiple first candidate emotion collocation groups, deletes and does not meet the first time for presetting interdependent rule
Selection sense collocation group.
In the present embodiment, text processing apparatus is in the first candidate emotion collocation for obtaining each existing user comment text
After group, the multiple first candidate emotion collocation groups can be screened according to preset interdependent rule, deletion do not meet it is default according to
Deposit the first candidate emotion collocation group of rule.
Specifically, interdependent rule is the evaluation object of the dependence and dimension word of confinement dimension word and evaluating word
The rule of dependence.Illustratively, the interdependent rule in the present embodiment are as follows: evaluating word in the first candidate emotion collocation group with
There is dynamic guest's relationship in dimension word, and/or, the dimension word in the first candidate emotion collocation group is corresponding with the first candidate emotion collocation group
Existing user comment text in comment object there are subject-predicate relationships.Text processing apparatus will not meet the interdependent rule
First candidate emotion collocation group is deleted.It is envisioned that other interdependent rules can also be used in those skilled in the art
The relationships such as the evaluation object of dependence and dimension word to dimension word and evaluating word in emotion collocation group constrain.
Illustratively, the first candidate emotion collocation group is " taste, pretty good ", and dimension word therein is " taste ", evaluation
Word is " pretty good ", is wherein guest's relationship between dimension word " taste " and evaluating word " pretty good ", and dimension word " taste " and
It is subject-predicate relationship between evaluation object " this family dining room ";The first candidate emotion collocation group meets this and presets interdependent rule.
S304 carries out feeling polarities analysis to each first candidate emotion collocation group, obtains each first candidate emotion and takes
The corresponding emotion word of combo.
Target emotion collocation group in the present embodiment further includes emotion word, wherein emotion word is existing user comment text
This feeling polarities, specifically, the feeling polarities in the present embodiment may include " positive " and " passiveness ".
Wherein, feeling polarities word library is previously stored in text processing apparatus, such as the word that feeling polarities are " positive "
Repertorie includes multiple words, and feeling polarities are that the word library of " passiveness " also includes multiple words;Specifically, feeling polarities word library
It is to use to have supervision, unsupervised or semi-supervised classification method, will will include the sample word progress polarity point of feeling polarities
Analysis, such as part of speech, morphology analysis are carried out, obtain feeling polarities word library.
Specifically, feeling polarities analysis is carried out to each first candidate emotion collocation group in the present embodiment, it can be using pre-
The feeling polarities word library first obtained classifies to the first candidate emotion collocation group, obtains each first candidate emotion collocation group
Corresponding emotion word.
Illustratively, the first candidate emotion collocation group is " taste, pretty good ", and corresponding emotion word is " positive ".
S305 obtains the emotion dictionary of collocations of target domain according to the multiple first candidate emotion collocation groups.
It, can after text processing apparatus obtains the corresponding emotion word of each first candidate emotion collocation group in the present embodiment
To obtain target domain according to each first candidate emotion collocation group and the corresponding emotion word of each first candidate emotion collocation group
Emotion dictionary of collocations.
Specifically, can be by each first candidate emotion collocation group and the corresponding emotion of each first candidate emotion collocation group
Word is combined, and obtains the second candidate emotion collocation group, each second candidate emotion collocation group includes dimension word, evaluating word and feelings
Feel word.
Illustratively, the first candidate emotion collocation group is " taste, pretty good ", and corresponding emotion word is " positive ", then
The corresponding second candidate emotion collocation group of first candidate emotion collocation group is " taste, pretty good, positive ".
After obtaining the corresponding second candidate emotion collocation group of each first candidate emotion collocation group, text processing apparatus root
It is semantic according to the first semantic and evaluating word second of the dimension word of each second candidate emotion collocation group, to identical first language
Justice and the second semantic second candidate emotion collocation group are clustered, and third candidate emotion collocation group is obtained.
Illustratively, as the second candidate emotion collocation group in " cuisines " field is " service, good, actively ", " attitude, very
It is good, positive ", due to being attendant, root in the dimension word " service " in " cuisines " field and the evaluation object of " attitude "
According to the first of dimension word semantic and evaluating word second semantic, this two second candidate emotion collocation group the first languages having the same
Therefore this two second candidate emotion collocation groups can be clustered, obtain the collocation of third candidate emotion by justice and the second semanteme
Group, third candidate's emotion collocation group after being clustered such as this two second candidate emotion collocation groups can for " service, it is good,
Actively ".
Further, in the present embodiment, after text processing apparatus clusters all second candidate emotion collocation group,
Obtain the target emotion collocation group of the target domain;Specifically, text processing apparatus is according to each third candidate emotion collocation group
The quantity of corresponding second candidate emotion collocation group, and be ranked up according to the sequence of quantity from big to small, it will be arranged in front pre-
If emotion dictionary of collocations of the third candidate emotion collocation group of quantity as target domain.
Illustratively, if third candidate's emotion collocation group " servicing, good, positive " is taken by 100 second candidate emotions
Combo carries out cluster acquisition, then the quantity of the corresponding second candidate emotion collocation group of third candidate emotion collocation group is 100,
It is preset with preset quantity in text processing apparatus, after being ranked up according to the sequence of quantity from big to small, will be arranged in front pre-
If emotion dictionary of collocations of the third candidate emotion collocation group of quantity as target domain.As it can be seen that the feelings that the present embodiment obtains
Feel the target emotion collocation group word frequency with higher in dictionary of collocations, there is preferable universality in the target domain.
In the present embodiment, text processing apparatus needs to obtain to be processed after the emotion dictionary of collocations for obtaining target domain
The corresponding emotion collocation group of user comment text, specifically, being divided into three kinds of situations in the present embodiment to how obtaining use to be processed
Comment text corresponding emotion collocation group in family is illustrated, and specifically includes S306-S308.
S306, if in user comment text to be processed including the dimension in emotion dictionary of collocations in first object emotion collocation group
Word and evaluating word are spent, then using first object emotion collocation group as the corresponding emotion collocation group of user comment text to be processed.
A method of obtaining the corresponding emotion collocation group of user comment text to be processed are as follows: in the feelings for obtaining target domain
It include the dimension in emotion dictionary of collocations in first object emotion collocation group after feeling dictionary of collocations, in user comment text to be processed
Word and evaluating word, then using first object emotion collocation group as the corresponding emotion collocation group of user comment text to be processed, wherein
First object emotion collocation group is any one target emotion collocation group in emotion dictionary of collocations.
Illustratively, target emotion collocation group is " service, is good, actively ", and user comment text to be processed is " this family
The service in dining room is all well and good, I is delithted with ", which includes the dimension word in target emotion collocation group
" service " and evaluating word " good " then takes " service, good, actively " as the corresponding emotion of the user comment text to be processed
Combo.
S307, if only including in the second target emotion collocation group in emotion dictionary of collocations in user comment text to be processed
Dimension word, and the feeling polarities of user comment text to be processed and the feeling polarities of the second target emotion collocation group are identical, then will
Second target emotion collocation group is as the corresponding emotion collocation group of user comment text to be processed.
Another kind obtains the mode of the corresponding emotion collocation group of user comment text to be processed are as follows: is obtaining target domain
It include the dimension in emotion dictionary of collocations in first object emotion collocation group after emotion dictionary of collocations, in user comment text to be processed
Word is spent, and the feeling polarities of user comment text to be processed and the feeling polarities of the second target emotion collocation group are identical, then by the
Two target emotion collocation groups are as the corresponding emotion collocation group of user comment text to be processed;Wherein, the second target emotion is arranged in pairs or groups
Group is any one target emotion collocation group in emotion dictionary of collocations.
Illustratively, target emotion collocation group is " service, is good, actively ", and user comment text to be processed is " this family
The service in dining room is very good, five-star level ", which includes the dimension word in target emotion collocation group
" service ", and the feeling polarities of the user comment text to be processed are the feeling polarities in " positive ", with target emotion collocation group
It is identical, thus can should " service, good, actively " as the corresponding emotion collocation group of the user comment text to be processed.
S308, if in user comment text to be processed not including any one target emotion collocation group in emotion dictionary of collocations
In dimension word, then by with the semantic similarity of user comment text to be processed be greater than similarity threshold target emotion collocation group
As the corresponding emotion collocation group of user comment text to be processed.
Another obtains the mode of the corresponding emotion collocation group of user comment text to be processed are as follows: is obtaining target domain
It does not include in emotion dictionary of collocations in first object emotion collocation group after emotion dictionary of collocations, in user comment text to be processed
Dimension word then obtains the semanteme of text to be processed and the semanteme of each target emotion collocation group, specifically, each target feelings
Feeling the semantic of collocation group can be first semantic and evaluating word the second semantic summation of dimension word therein.
Text processing apparatus states user comment text to be processed using the method acquisition of implicit semantic computation in the present embodiment
With the semantic similarity of each target emotion collocation group, wherein learn implicit semantic computation method may include word2vec,
The modes such as phrase2vec, CNN, LSTM are not specifically limited in the present embodiment.
Specifically, text processing apparatus will be greater than similarity threshold with the semantic similarity of user comment text to be processed
Target emotion collocation group is as the corresponding emotion collocation group of user comment text to be processed;It is envisioned that if more than similar
It, can be using the corresponding target emotion collocation group of maximum value similarity as wait locate when the target emotion collocation group of degree threshold value is multiple
Manage the corresponding emotion collocation group of user comment text.
S306-S308 in the present embodiment does not have the difference of sequencing, and three is three kinds of sides independently implemented
Case.
In the present embodiment, word segmentation processing is carried out to each existing user comment text, each existing user is obtained and comments
Multiple words of paper sheet, further according to the corresponding part of speech of multiple words of each existing user comment text, and, part of speech is taken
With rule, the first candidate emotion collocation group of each existing user comment text is obtained, and each first candidate emotion is taken
Combo carries out feeling polarities analysis, obtains the corresponding emotion word of each first candidate emotion collocation group, by each first candidate feelings
Sense collocation group and the corresponding emotion word of each first candidate emotion collocation group are combined, and obtain the second candidate emotion collocation group;
Further, semantic according to the first semantic and evaluating word second of the second candidate emotion collocation group dimension word, to the second candidate
Emotion collocation group is clustered, and is obtained the emotion dictionary of collocations of target domain, is avoided artificial constructed emotion in the present embodiment and take
With a large amount of manpower of waste caused by dictionary, the problem of low efficiency;And include in user comment text to be processed in the present embodiment
Dimension word or dimension word and evaluating word or user comment text to be processed in emotion collocation group do not include emotion collocation group
In dimension word when, obtain the corresponding emotion collocation group of user comment text to be processed respectively, can accurately obtain target domain
Text emotion viewpoint.
Fig. 4 is the structural schematic diagram one of text processing apparatus provided by the invention, as shown in figure 4, text processing unit
400 include: that emotion dictionary of collocations obtains module 401 and emotion collocation group acquisition module 402.
Emotion dictionary of collocations obtains module 401, for the existing user comment text according to target domain, obtains target
The emotion dictionary of collocations in field, includes multiple target emotion collocation groups in emotion dictionary of collocations, and each target emotion collocation group is used
It is commented in emotion of the characterization user to the attribute of the comment object of target domain.
Emotion collocation group obtains module 402, for being arranged in pairs or groups according to the user comment text to be processed and emotion of target domain
Dictionary obtains the corresponding emotion collocation group of user comment text to be processed.
Text processing apparatus provided in this embodiment is similar with principle and technical effect that above-mentioned text handling method is realized,
Therefore not to repeat here.
Optionally, Fig. 5 is the structural schematic diagram two of text processing apparatus provided by the invention, as shown in figure 5, at the text
Manage device 400 further include: emotion word obtains module 403 and removing module 404.
Emotion word obtains module 403, for carrying out feeling polarities analysis to each first candidate emotion collocation group, obtains every
The corresponding emotion word of a first candidate emotion collocation group.
Removing module 404, for screening to the multiple first candidate emotion collocation groups, deletion, which is not met, presets interdependent rule
The candidate emotion collocation group of first then, presets interdependent rule are as follows: the evaluating word in the first candidate emotion collocation group is deposited with dimension word
In dynamic guest's relationship, and/or, the dimension word and the first candidate emotion collocation group in the first candidate emotion collocation group are corresponding existing
There are subject-predicate relationships for comment object in user comment text.
Optionally, target emotion collocation group includes dimension word and evaluating word, and dimension word is in existing user comment text
Comment object attribute.
Optionally, emotion dictionary of collocations obtains module 401, specifically for dividing each existing user comment text
Word processing, obtains multiple words of each existing user comment text;According to the multiple of each existing user comment text
The corresponding part of speech of word, and, part of speech collocation rule obtains the first candidate emotion collocation of each existing user comment text
Group, part of speech collocation rule includes: that dimension word is established rules really, and evaluating word is established rules then really;It is taken according to the multiple first candidate emotions
Combo obtains the emotion dictionary of collocations of target domain.
Optionally, target emotion collocation group further includes emotion word, and emotion word is the emotion pole of existing user comment text
Property.
Optionally, emotion dictionary of collocations obtains module 401, specifically for according to each first candidate emotion collocation group and often
The corresponding emotion word of a first candidate emotion collocation group, obtains the emotion dictionary of collocations of target domain.
Optionally, emotion dictionary of collocations obtains module 401, specifically for by each first candidate emotion collocation group and each
The corresponding emotion word of first candidate's emotion collocation group is combined, and obtains the second candidate emotion collocation group, each second candidate feelings
Feeling collocation group includes dimension word, evaluating word and emotion word;According to the first language of the dimension word of each second candidate emotion collocation group
Second semanteme of justice and evaluating word gathers to identical first semantic and the second semanteme the second candidate emotion collocation group
Class obtains third candidate emotion collocation group;According to the corresponding second candidate emotion collocation group of each third candidate emotion collocation group
Quantity, and be ranked up according to the sequence of quantity from big to small, the third candidate's emotion that will be arranged in front preset quantity is taken
Emotion dictionary of collocations of the combo as target domain.
Optionally, emotion collocation group obtains module 402, if specifically for taking in user comment text to be processed including emotion
With the dimension word and evaluating word in dictionary in first object emotion collocation group, then using first object emotion collocation group as to be processed
The corresponding emotion collocation group of user comment text, first object emotion collocation group are any one target in emotion dictionary of collocations
Emotion collocation group;If in user comment text to be processed only including the dimension in emotion dictionary of collocations in the second target emotion collocation group
Word is spent, and the feeling polarities of user comment text to be processed and the feeling polarities of the second target emotion collocation group are identical, then by the
For two target emotion collocation groups as the corresponding emotion collocation group of user comment text to be processed, the second target emotion collocation group is feelings
Feel any one target emotion collocation group in dictionary of collocations.
Optionally, emotion collocation group obtains module 402, if specifically for not including emotion in user comment text to be processed
Dimension word in dictionary of collocations in any one target emotion collocation group, then will be similar to the semanteme of user comment text to be processed
Degree is greater than the target emotion collocation group of similarity threshold as the corresponding emotion collocation group of user comment text to be processed.
Fig. 6 is the structural schematic diagram three of text processing apparatus provided by the invention, and text processing unit for example can be
Terminal device, such as smart phone, tablet computer, computer etc..As shown in fig. 6, text processing unit 600 includes: storage
Device 601 and at least one processor 602.
Memory 601, for storing program instruction.
Processor 602, for being performed the text handling method realized in the present embodiment, specific implementation in program instruction
Principle can be found in above-described embodiment, and details are not described herein again for the present embodiment.
Text processing unit 600 can also include and input/output interface 603.
Input/output interface 603 may include independent output interface and input interface, or integrated input and defeated
Integrated interface out.Wherein, output interface is used for output data, and input interface is used to obtain the data of input, above-mentioned output
Data are the general designation exported in above method embodiment, and the data of input are the general designation inputted in above method embodiment.
The present invention also provides a kind of readable storage medium storing program for executing, it is stored with and executes instruction in readable storage medium storing program for executing, work as text-processing
When at least one processor of device executes this and executes instruction, when computer executed instructions are executed by processor, realize above-mentioned
Text handling method in embodiment.
The present invention also provides a kind of program product, the program product include execute instruction, this execute instruction be stored in it is readable
In storage medium.At least one processor of text processing apparatus can read this from readable storage medium storing program for executing and execute instruction, at least
One processor executes this and executes instruction so that text processing apparatus implements the text-processing that above-mentioned various embodiments provide
Method.
In several embodiments provided by the present invention, it should be understood that disclosed device and method can pass through it
Its mode is realized.For example, the apparatus embodiments described above are merely exemplary, for example, the division of the unit, only
Only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple units or components can be tied
Another system is closed or is desirably integrated into, or some features can be ignored or not executed.Another point, it is shown or discussed
Mutual coupling, direct-coupling or communication connection can be through some interfaces, the INDIRECT COUPLING or logical of device or unit
Letter connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of hardware adds SFU software functional unit.
The above-mentioned integrated unit being realized in the form of SFU software functional unit can store and computer-readable deposit at one
In storage media.Above-mentioned SFU software functional unit is stored in a storage medium, including some instructions are used so that a computer
Equipment (can be personal computer, server or the network equipment etc.) or processor (English: processor) execute this hair
The part steps of bright each embodiment the method.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory
(English: Read-Only Memory, abbreviation: ROM), random access memory (English: Random Access Memory, letter
Claim: RAM), the various media that can store program code such as magnetic or disk.
In the embodiment of the above-mentioned network equipment or terminal device, it should be appreciated that processor can be central processing unit
(English: Central Processing Unit, referred to as: CPU), it can also be other general processors, digital signal processor
(English: Digital Signal Processor, abbreviation: DSP), specific integrated circuit (English: Application
Specific Integrated Circuit, referred to as: ASIC) etc..General processor can be microprocessor or the processor
It is also possible to any conventional processor etc..Hardware handles can be embodied directly in conjunction with the step of method disclosed in the present application
Device executes completion, or in processor hardware and software module combination execute completion.
Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention., rather than its limitations;To the greatest extent
Pipe present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: its according to
So be possible to modify the technical solutions described in the foregoing embodiments, or to some or all of the technical features into
Row equivalent replacement;And these are modified or replaceed, various embodiments of the present invention technology that it does not separate the essence of the corresponding technical solution
The range of scheme.
Claims (10)
1. a kind of text handling method characterized by comprising
According to the existing user comment text of target domain, the emotion dictionary of collocations of the target domain, the emotion are obtained
It include multiple target emotion collocation groups in dictionary of collocations, each target emotion collocation group is for characterizing user to the target domain
Comment object attribute emotion comment;
According to the user comment text to be processed of the target domain and the emotion dictionary of collocations, the user to be processed is obtained
The corresponding emotion collocation group of comment text.
2. the method according to claim 1, wherein the target emotion collocation group includes dimension word and evaluation
Word, the dimension word are the attribute of the comment object in the existing user comment text;It is described according to target domain
Some user comment texts obtain the emotion dictionary of collocations of the target domain, comprising:
Word segmentation processing is carried out to each existing user comment text, obtains each existing user comment text
Multiple words;
According to the corresponding part of speech of multiple words of each existing user comment text, and, part of speech collocation rule obtains
The candidate emotion collocation group of the first of each existing user comment text, the part of speech collocation rule includes: dimension word
Determine that rule and evaluating word are established rules then really;
According to the multiple described first candidate emotion collocation groups, the emotion dictionary of collocations of the target domain is obtained.
3. described according to the method described in claim 2, it is characterized in that, the target emotion collocation group further includes emotion word
Emotion word is the feeling polarities of the existing user comment text;The each existing user comment text of acquisition
After first candidate emotion collocation group, further includes:
Feeling polarities analysis is carried out to each described first candidate emotion collocation group, obtains each described first candidate emotion collocation
The corresponding emotion word of group;
It is described according to multiple first candidate emotion collocation groups, obtain the emotion dictionary of collocations of the target domain, comprising:
According to each described first candidate emotion collocation group and the corresponding emotion word of each first candidate emotion collocation group, obtain
Take the emotion dictionary of collocations of the target domain.
4. according to the method described in claim 3, it is characterized in that, it is described according to each first candidate emotion collocation group and
The corresponding emotion word of each first candidate emotion collocation group, obtains the emotion dictionary of collocations of the target domain, comprising:
Each described first candidate emotion collocation group and the corresponding emotion word of each first candidate emotion collocation group are carried out
Combination obtains the second candidate emotion collocation group, and each described second candidate emotion collocation group includes the dimension word, the evaluation
Word and the emotion word;
It is semantic according to the first semantic and evaluating word second of the dimension word of each second candidate emotion collocation group, to having
Identical first semantic and the second semanteme the second candidate emotion collocation group is clustered, and third candidate emotion collocation group is obtained;
According to the quantity of the corresponding second candidate emotion collocation group of each third candidate emotion collocation group, and according to the number
The sequence of amount from big to small is ranked up, and will be arranged in front third candidate's emotion collocation group of preset quantity as the target
The emotion dictionary of collocations in field.
5. according to the method described in claim 3, it is characterized in that, described carry out each described first candidate emotion collocation group
Feeling polarities are analyzed, before the corresponding emotion word of the candidate emotion collocation group of acquisition each described first, further includes:
Multiple described first candidate emotion collocation groups are screened, deletes and does not meet the first candidate emotion for presetting interdependent rule
Collocation group, it is described to preset interdependent rule are as follows: evaluating word and dimension word in the described first candidate emotion collocation group have dynamic guest and close
System, and/or, the dimension word and the described first candidate emotion collocation group in the described first candidate emotion collocation group are corresponding existing
There are subject-predicate relationships for comment object in user comment text.
6. according to the described in any item methods of claim 2-4, which is characterized in that described according to the to be processed of the target domain
User comment text and the emotion dictionary of collocations obtain the corresponding emotion collocation group of the user comment text to be processed, packet
It includes:
If in the user comment text to be processed including the dimension in the emotion dictionary of collocations in first object emotion collocation group
Word and evaluating word are spent, then is taken the first object emotion collocation group as the corresponding emotion of the user comment text to be processed
Combo, the first object emotion collocation group are any one target emotion collocation group in the emotion dictionary of collocations;
If only including in the second target emotion collocation group in the emotion dictionary of collocations in the user comment text to be processed
Dimension word, and the feeling polarities phase of the feeling polarities of the user comment text to be processed and the second target emotion collocation group
Together, then described using the second target emotion collocation group as the corresponding emotion collocation group of the user comment text to be processed
Second target emotion collocation group is any one target emotion collocation group in the emotion dictionary of collocations.
7. according to the described in any item methods of claim 2-4, which is characterized in that described according to the to be processed of the target domain
User comment text and the emotion dictionary of collocations obtain the corresponding emotion collocation group of the user comment text to be processed, packet
It includes:
If in the user comment text to be processed not including any one target emotion collocation group in the emotion dictionary of collocations
In dimension word, then by with the semantic similarity of the user comment text to be processed be greater than similarity threshold target emotion take
Combo is as the corresponding emotion collocation group of the user comment text to be processed.
8. a kind of text processing apparatus characterized by comprising
Emotion dictionary of collocations obtains module, for the existing user comment text according to target domain, obtains the target neck
The emotion dictionary of collocations in domain includes multiple target emotion collocation groups, each target emotion collocation group in the emotion dictionary of collocations
The emotion of the attribute of the comment object of the target domain is commented on for characterizing user;
Emotion collocation group obtains module, for according to the user comment text to be processed of the target domain and emotion collocation
Dictionary obtains the corresponding emotion collocation group of the user comment text to be processed.
9. a kind of text processing apparatus characterized by comprising at least one processor and memory;
The memory stores computer executed instructions;
At least one described processor executes the computer executed instructions of the memory storage, so that the text processing apparatus
Perform claim requires the described in any item methods of 1-7.
10. a kind of computer readable storage medium, which is characterized in that be stored with computer on the computer readable storage medium
It executes instruction, when the computer executed instructions are executed by processor, realizes the described in any item methods of claim 1-7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811539984.5A CN109800418B (en) | 2018-12-17 | 2018-12-17 | Text processing method, device and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811539984.5A CN109800418B (en) | 2018-12-17 | 2018-12-17 | Text processing method, device and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109800418A true CN109800418A (en) | 2019-05-24 |
CN109800418B CN109800418B (en) | 2023-05-05 |
Family
ID=66556976
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811539984.5A Active CN109800418B (en) | 2018-12-17 | 2018-12-17 | Text processing method, device and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109800418B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111144507A (en) * | 2019-12-30 | 2020-05-12 | 北京百度网讯科技有限公司 | Emotion analysis model pre-training method and device and electronic equipment |
CN111191428A (en) * | 2019-12-27 | 2020-05-22 | 北京百度网讯科技有限公司 | Comment information processing method and device, computer equipment and medium |
CN111209371A (en) * | 2019-12-31 | 2020-05-29 | 新华网股份有限公司 | Comment data processing method and device, computer equipment and storage medium |
CN111832313A (en) * | 2020-06-09 | 2020-10-27 | 北京百度网讯科技有限公司 | Method, device, equipment and medium for generating emotion collocation set in text |
CN116738298A (en) * | 2023-08-16 | 2023-09-12 | 杭州同花顺数据开发有限公司 | Text classification method, system and storage medium |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080133488A1 (en) * | 2006-11-22 | 2008-06-05 | Nagaraju Bandaru | Method and system for analyzing user-generated content |
CN104317965A (en) * | 2014-11-14 | 2015-01-28 | 南京理工大学 | Establishment method of emotion dictionary based on linguistic data |
CN105868185A (en) * | 2016-05-16 | 2016-08-17 | 南京邮电大学 | Part-of-speech-tagging-based dictionary construction method applied in shopping comment emotion analysis |
CN105989550A (en) * | 2015-03-05 | 2016-10-05 | 阿里巴巴集团控股有限公司 | Online service evaluation information determination method and equipment |
CN106776574A (en) * | 2016-12-28 | 2017-05-31 | Tcl集团股份有限公司 | User comment text method for digging and device |
US20170249384A1 (en) * | 2016-02-29 | 2017-08-31 | Microsoft Technology Licensing, Llc | Content categorization |
CN107203520A (en) * | 2016-03-16 | 2017-09-26 | 中国科学院上海高等研究院 | The method for building up of hotel's sentiment dictionary, the sentiment analysis method and system of comment |
CN107291696A (en) * | 2017-06-28 | 2017-10-24 | 达而观信息科技(上海)有限公司 | A kind of comment word sentiment analysis method and system based on deep learning |
CN107832297A (en) * | 2017-11-09 | 2018-03-23 | 电子科技大学 | A kind of field sentiment dictionary construction method of Feature Oriented word granularity |
CN108763214A (en) * | 2018-05-30 | 2018-11-06 | 河海大学 | A kind of sentiment dictionary method for auto constructing for comment on commodity |
-
2018
- 2018-12-17 CN CN201811539984.5A patent/CN109800418B/en active Active
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080133488A1 (en) * | 2006-11-22 | 2008-06-05 | Nagaraju Bandaru | Method and system for analyzing user-generated content |
CN104317965A (en) * | 2014-11-14 | 2015-01-28 | 南京理工大学 | Establishment method of emotion dictionary based on linguistic data |
CN105989550A (en) * | 2015-03-05 | 2016-10-05 | 阿里巴巴集团控股有限公司 | Online service evaluation information determination method and equipment |
US20170249384A1 (en) * | 2016-02-29 | 2017-08-31 | Microsoft Technology Licensing, Llc | Content categorization |
CN107203520A (en) * | 2016-03-16 | 2017-09-26 | 中国科学院上海高等研究院 | The method for building up of hotel's sentiment dictionary, the sentiment analysis method and system of comment |
CN105868185A (en) * | 2016-05-16 | 2016-08-17 | 南京邮电大学 | Part-of-speech-tagging-based dictionary construction method applied in shopping comment emotion analysis |
CN106776574A (en) * | 2016-12-28 | 2017-05-31 | Tcl集团股份有限公司 | User comment text method for digging and device |
CN107291696A (en) * | 2017-06-28 | 2017-10-24 | 达而观信息科技(上海)有限公司 | A kind of comment word sentiment analysis method and system based on deep learning |
CN107832297A (en) * | 2017-11-09 | 2018-03-23 | 电子科技大学 | A kind of field sentiment dictionary construction method of Feature Oriented word granularity |
CN108763214A (en) * | 2018-05-30 | 2018-11-06 | 河海大学 | A kind of sentiment dictionary method for auto constructing for comment on commodity |
Non-Patent Citations (3)
Title |
---|
YICHAO REN ETC.: "Sentiment analysis of Internet performance data", 《2017 IEEE 3RD INFORMATION TECHNOLOGY AND MECHATRONICS ENGINEERING CONFERENCE (ITOEC)》 * |
孔伟俊 等: "基于领域词典的网络商品评论情感分析", 《计算机与数字工程》 * |
李耀林: "面向评价对象的商品评论情感倾向性分析研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111191428A (en) * | 2019-12-27 | 2020-05-22 | 北京百度网讯科技有限公司 | Comment information processing method and device, computer equipment and medium |
CN111191428B (en) * | 2019-12-27 | 2022-02-25 | 北京百度网讯科技有限公司 | Comment information processing method and device, computer equipment and medium |
US11507751B2 (en) | 2019-12-27 | 2022-11-22 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Comment information processing method and apparatus, and medium |
CN111144507A (en) * | 2019-12-30 | 2020-05-12 | 北京百度网讯科技有限公司 | Emotion analysis model pre-training method and device and electronic equipment |
US11537792B2 (en) | 2019-12-30 | 2022-12-27 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Pre-training method for sentiment analysis model, and electronic device |
CN111209371A (en) * | 2019-12-31 | 2020-05-29 | 新华网股份有限公司 | Comment data processing method and device, computer equipment and storage medium |
CN111209371B (en) * | 2019-12-31 | 2024-06-07 | 新华网股份有限公司 | Comment data processing method, comment data processing device, computer equipment and storage medium |
CN111832313A (en) * | 2020-06-09 | 2020-10-27 | 北京百度网讯科技有限公司 | Method, device, equipment and medium for generating emotion collocation set in text |
CN111832313B (en) * | 2020-06-09 | 2023-07-25 | 北京百度网讯科技有限公司 | Method, device, equipment and medium for generating emotion matching set in text |
CN116738298A (en) * | 2023-08-16 | 2023-09-12 | 杭州同花顺数据开发有限公司 | Text classification method, system and storage medium |
CN116738298B (en) * | 2023-08-16 | 2023-11-24 | 杭州同花顺数据开发有限公司 | Text classification method, system and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN109800418B (en) | 2023-05-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109800418A (en) | Text handling method, device and storage medium | |
CN106156365B (en) | A kind of generation method and device of knowledge mapping | |
CN104462363B (en) | Comment point shows method and apparatus | |
JP5534280B2 (en) | Text clustering apparatus, text clustering method, and program | |
CN105095288B (en) | Data analysis method and data analysis device | |
CN107729468A (en) | Answer extracting method and system based on deep learning | |
List et al. | Using phylogenetic networks to model Chinese dialect history | |
CN107301170A (en) | The method and apparatus of cutting sentence based on artificial intelligence | |
WO2015053236A1 (en) | Device for collecting contradictory expression and computer program for same | |
JP2013175097A (en) | Predicate template gathering device, specific phrase pair gathering device, and computer program therefor | |
CN112836487B (en) | Automatic comment method and device, computer equipment and storage medium | |
CN106599110A (en) | Artificial intelligence-based voice search method and device | |
Maru et al. | Nibbling at the hard core of Word Sense Disambiguation | |
Bellomaria et al. | Almawave-SLU: A new dataset for SLU in Italian | |
CN111125360A (en) | Emotion analysis method and device in game field and model training method and device thereof | |
CN106569996B (en) | A kind of Sentiment orientation analysis method towards Chinese microblogging | |
JP5591871B2 (en) | Answer type estimation apparatus, method, and program | |
CN109992665A (en) | A kind of classification method based on the extension of problem target signature | |
JP6420268B2 (en) | Image evaluation learning device, image evaluation device, image search device, image evaluation learning method, image evaluation method, image search method, and program | |
CN111428503A (en) | Method and device for identifying and processing same-name person | |
CN109284389A (en) | A kind of information processing method of text data, device | |
Fuad et al. | Analysis and classification of mobile apps using topic modeling: A case study on Google Play Arabic apps | |
CN110069686A (en) | User behavior analysis method, apparatus, computer installation and storage medium | |
CN108460016A (en) | A kind of entity name analysis recognition method | |
CN109241438B (en) | Element-based cross-channel hot event discovery method and device and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |