CN110472018A - Information processing method, device and computer storage medium based on deep learning - Google Patents
Information processing method, device and computer storage medium based on deep learning Download PDFInfo
- Publication number
- CN110472018A CN110472018A CN201910776373.0A CN201910776373A CN110472018A CN 110472018 A CN110472018 A CN 110472018A CN 201910776373 A CN201910776373 A CN 201910776373A CN 110472018 A CN110472018 A CN 110472018A
- Authority
- CN
- China
- Prior art keywords
- information
- search word
- searched
- degree
- word
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/3332—Query translation
- G06F16/3338—Query expansion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3347—Query execution using vector based model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
It includes: to obtain the keyword entity information of information to be searched according to Information Extraction Model that the embodiment of the present application, which provides a kind of information processing method based on deep learning, device and computer storage medium, method,;According to the keyword entity information of the information to be searched and relative positional relationship of the target search word in multi-C vector space, the degree of correlation of the information to be searched Yu the target search word is determined;Determine the information to be searched to the disturbance degree of the target search word according to emotion model;The information to be searched is ranked up according to the degree of correlation and the disturbance degree, obtains the degree of correlation corresponding with the target search word and the maximum information to be searched of disturbance degree;The application can quickly, accurately and reliably be filtered massive information, pluck choosing, understanding and matching treatment.
Description
Technical field
This application involves deep learning fields, and in particular to a kind of information processing method based on deep learning, device and
Computer storage medium.
Background technique
As universal and development, the more and more information of internet and mobile Internet are disclosed and are sent out on the internet
Ferment is obtained to the valuable information of enterprise current particularly important by the data of magnanimity on detection internet, such as passes through this
A little information enterprises can be found that collage-credit data and enterprise's public sentiment.
Internet data information is monitored, the scheme used in currently available technology is magnanimity crawl internet data
Afterwards, acquisition valid data are filtered by modes such as manual analysis or keyword searches, for example current mainstream information monitoring is put down
Platform such as Sina's public sentiment is logical and Enterprises'Forewarning is logical, and the mode that keyword filtering can be used in these platforms collects valid data.
Inventors have found that solution in the prior art stores following defect and deficiency:
1, it is filtered according to keyword, accuracy rate is low, and coverage rate is poor.It is Chinese of extensive knowledge and profound scholarship, very by setting keyword
The content of hardly possible covering institute collection in need, describes to emerge one after another, and keyword is in sentence for the keyword of object content
Effect can produce ambiguity, for example target keyword is " Jingdone district ", and actual content may be " in Beijing Dongcheng District " that can see
Such case can reduce accuracy rate out;
2, further screening and sequence are carried out to filtered information according to the mode of manual read, and identification information is made
At influence coverage rate it is poor, low efficiency, for the internet for being flooded with mass data, the mode efficiency of artificial filter is too
Low, although the mode filtered using keyword, data analyst still will face the information list of several louvers daily, browse institute
Some contents are nearly impossible.
Summary of the invention
For the problems of the prior art, the application provide a kind of information processing method based on deep learning, device and
Computer storage medium can quickly, accurately and reliably be filtered massive information, pluck choosing, understanding and matching treatment.
At least one of to solve the above-mentioned problems, the application the following technical schemes are provided:
In a first aspect, the application provides a kind of information processing method based on deep learning, comprising:
The keyword entity information of information to be searched is obtained according to Information Extraction Model;
It is opposite in multi-C vector space according to the keyword entity information of the information to be searched and target search word
Positional relationship determines the degree of correlation of the information to be searched Yu the target search word;
Determine the information to be searched to the disturbance degree of the target search word according to emotion model;
The information to be searched is ranked up according to the degree of correlation and the disturbance degree, is obtained and the target search
The corresponding degree of correlation of word and the maximum information to be searched of disturbance degree.
Further, the keyword entity information according to the information to be searched and target search word multidimensional to
Relative positional relationship in quantity space, before the degree of correlation for determining the information to be searched and the target search word, comprising:
According to first position of the initial ranging word of user's input in multi-C vector space, obtain and the first position
Distance meet the expanded search word of default neighbor distance condition;
The initial ranging word and the expanded search word are set as the target search word.
Further, it is described by the initial ranging word and the expanded search word be set as the target search word it
Before, comprising:
By the initial ranging word inputted with user, corresponding first instance meets default close in default knowledge mapping
The second instance of connection relationship is set as supplementing expanded search word;
Vocabulary supplement and vocabulary is carried out to the expanded search word according to the supplement expanded search word to optimize, obtain by
The expanded search word after vocabulary supplement and vocabulary optimization.
Further, the first position described according to the initial ranging word of user's input in multi-C vector space, obtains
To before the expanded search word for meeting default neighbor distance condition at a distance from the first position, comprising:
Position of each search term in multi-C vector space is determined according to default participle model.
Second aspect, the application provide a kind of information processing unit based on deep learning, comprising:
Keyword abstraction module, for obtaining the keyword entity information of information to be searched according to Information Extraction Model;
Degree of correlation determining module, for the keyword entity information according to the information to be searched with target search word more
Relative positional relationship in dimensional vector space determines the degree of correlation of the information to be searched Yu the target search word;
Disturbance degree determining module, for determining the information to be searched to the shadow of the target search word according to emotion model
Loudness;
Information determination module to be searched, for being carried out according to the degree of correlation and the disturbance degree to the information to be searched
Sequence, obtains the degree of correlation corresponding with the target search word and the maximum information to be searched of disturbance degree.
Further, further includes:
Expanded search word determination unit, initial ranging word for being inputted according to user in multi-C vector space first
Position obtains the expanded search word for meeting default neighbor distance condition at a distance from the first position;
Target search word determination unit, for the initial ranging word and the expanded search word to be set as the target
Search term.
Further, further includes:
Expansion word determination unit is supplemented, the initial ranging word for will input with user is right in default knowledge mapping
The second instance that the first instance answered meets preset association relationship is set as supplementing expanded search word;
Supplement optimization unit, for according to the supplement expanded search word to the expanded search word carry out vocabulary supplement with
Vocabulary optimization obtains the expanded search word after vocabulary supplements and vocabulary optimizes.
Further, further includes:
Position map unit, for determining position of each search term in multi-C vector space according to default participle model
It sets.
The third aspect, the application provides a kind of electronic equipment, including memory, processor and storage are on a memory and can
The computer program run on a processor, the processor realize the letter based on deep learning when executing described program
The step of ceasing processing method.
Fourth aspect, the application provide a kind of computer readable storage medium, are stored thereon with computer program, the calculating
The step of information processing method based on deep learning is realized when machine program is executed by processor.
As shown from the above technical solution, the application provide it is a kind of by the information processing method of deep learning, device and based on
Calculation machine storage medium obtains the keyword entity information of information to be searched by Information Extraction Model, and according to described to be searched
The keyword entity information and relative positional relationship of the target search word in multi-C vector space of information, determine described to be searched
Then the degree of correlation of information and the target search word determines the information to be searched to the target search according to emotion model
The disturbance degree of word is ranked up the information to be searched further according to the degree of correlation and the disturbance degree, obtains and the mesh
The corresponding degree of correlation of search term and the maximum information to be searched of disturbance degree are marked, the application passes through Information Extraction Model and emotion
Model is automatically ranked up the degree of correlation and disturbance degree of information to be searched, can quickly, accurately and reliably obtain the degree of correlation
Highest influences most important information, and can save a large amount of artificial investment, at the same when artificial treatment can also be reduced occur it is wrong
Probability accidentally.
Detailed description of the invention
In order to illustrate the technical solutions in the embodiments of the present application or in the prior art more clearly, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is the application
Some embodiments for those of ordinary skill in the art without creative efforts, can also basis
These attached drawings obtain other attached drawings.
Fig. 1 is one of the flow diagram of information processing method based on deep learning in the embodiment of the present application;
Fig. 2 is the two of the flow diagram of the information processing method based on deep learning in the embodiment of the present application;
Fig. 3 is the three of the flow diagram of the information processing method based on deep learning in the embodiment of the present application;
Fig. 4 is one of the structural schematic diagram of information processing unit based on deep learning in the embodiment of the present application;
Fig. 5 is the second structural representation of the information processing unit based on deep learning in the embodiment of the present application;
Fig. 6 is the third structural representation of the information processing unit based on deep learning in the embodiment of the present application;
Fig. 7 is the structural schematic diagram of the electronic equipment in the embodiment of the present application.
Specific embodiment
To keep the purposes, technical schemes and advantages of the embodiment of the present application clearer, below in conjunction with the embodiment of the present application
In attached drawing, technical solutions in the embodiments of the present application carries out clear, complete description, it is clear that described embodiment is
Some embodiments of the present application, instead of all the embodiments.Based on the embodiment in the application, those of ordinary skill in the art
Every other embodiment obtained without creative efforts, shall fall in the protection scope of this application.
In view of solution in the prior art stores following defect and deficiency: 1, it according to keyword is filtered, it is quasi-
True rate is low, and coverage rate is poor.It is Chinese of extensive knowledge and profound scholarship, by set keyword be difficult covering collection in need content, for mesh
The keyword description of mark content emerges one after another, and effect of the keyword in sentence can produce ambiguity, such as target keyword
It is " Jingdone district " that actual content may be " in Beijing Dongcheng District " it can be seen that such case can reduce accuracy rate;2, according to people
The mode that work is read carries out further screening and sequence, and influence coverage rate caused by identification information to filtered information
Difference, low efficiency, for the internet for being flooded with mass data, the too low problem of the mode efficiency of artificial filter, the application
A kind of information processing method based on deep learning, device and computer storage medium are provided, obtained by Information Extraction Model
The keyword entity information of information to be searched, and existed according to the keyword entity information and target search word of the information to be searched
Relative positional relationship in multi-C vector space determines the degree of correlation of the information to be searched Yu the target search word, then
The information to be searched is determined to the disturbance degree of the target search word, further according to the degree of correlation and described according to emotion model
Disturbance degree is ranked up the information to be searched, obtains the degree of correlation corresponding with the target search word and disturbance degree is maximum
The information to be searched, the application is by Information Extraction Model and emotion model automatically on the degree of correlation of information to be searched and influence
Degree is ranked up, and can quickly, accurately and reliably obtain degree of correlation highest influences most important information, and can be saved a large amount of
Artificial investment, while, there is the probability of mistake in can also reduce artificial treatment when.
In order to quickly, be accurately and reliably filtered to massive information, pluck choosing, understanding and matching treatment, this Shen
A kind of embodiment of information processing method based on deep learning please be provide, referring to Fig. 1, at the information based on deep learning
Reason method specifically includes following content:
Step S101: the keyword entity information of information to be searched is obtained according to Information Extraction Model.
Step S102: according to the keyword entity information of the information to be searched and target search word in multi-C vector space
In relative positional relationship, determine the degree of correlation of the information to be searched Yu the target search word.
Step S103: determine the information to be searched to the disturbance degree of the target search word according to emotion model.
Step S104: being ranked up the information to be searched according to the degree of correlation and the disturbance degree, obtains and institute
State the corresponding degree of correlation of target search word and the maximum information to be searched of disturbance degree.
It is understood that the Information Extraction Model can be a kind of existing keyword extraction model, such as can be with
By combining existing NER model and Entity-Relationship model to carry out information extraction, the Information Extraction Model
Training data can for pass through read article, data engineering teacher summarize article lists of keywords, using these training datas
It trains and obtains Information Extraction Model described herein.By Information Extraction Model described herein, can obtain described
The keyword entity information of information (such as input article) to be searched, such as information to be searched is Alibaba, listed company
The positive negative press of (code BABA), and the keyword entity information searched for is Taobao, and true with the Information Extraction Model
Positional relationship of the fixed keyword entity information and target search word in multi-C vector space, wherein the target search word can
To include the initial ranging word of user's input and with the initial ranging word, position is adjacent or close in the multi-C vector space
Expanded search word, and then obtain the degree of correlation of target search word Yu information to be searched, it is to be understood that the degree of correlation can not
It is only determined by the distance between position, can also include the sparse degree of the point in space around certain point.Such as put it is very dense, then
Only looking for immediate n point, (n can be set to any value, such as 300), if fruit dot is relatively sparse, then expands search range and looks for
To the point in the same cluster, Kmeans can be used to define each cluster.
It is understood that the emotion model can be a kind of model of existing determination correlation degree between the two,
Such as existing Deep learning based Sentimental Analysis model, the training data of the emotion model
It can be to be given a mark by the emotion manually to preset article and article, it can be to described wait search by the emotion model
Rope information carries out emotion prediction, and then obtains information to be searched to the disturbance degree of the target search word, it is to be understood that tool
Body disturbance degree is determined according to the NLP algorithm of deep learning, there is several factors decision, such as quantity, the phase of related term
The frequency, the influence degree of related content of the word in all words are closed, such as the disturbance degree that senior executive resigns just permission occurs than product
It is high.
It is understood that after getting the above-mentioned degree of correlation and disturbance degree data, it can be to corresponding described to be searched
Information is ranked up, such as carry out sequence sequence according to the degree of correlation, or according to disturbance degree carry out sequence sequence, and can be to the degree of correlation
Corresponding weighted value is preset with each ranking of disturbance degree, to obtain the degree of correlation weight of a certain information to be searched and influence
The sum of weight is spent, and then determines (such as degree of correlation weight and disturbance degree weight corresponding with the target search word, most important
The sum of it is maximum) information to be searched.
As can be seen from the above description, the information processing method provided by the embodiments of the present application based on deep learning, can pass through
Information Extraction Model obtains the keyword entity information of information to be searched, and is believed according to the keyword entity of the information to be searched
Breath and relative positional relationship of the target search word in multi-C vector space, determine the information to be searched and the target search
The degree of correlation of word, then according to emotion model determine the information to be searched to the disturbance degree of the target search word, further according to
The degree of correlation and the disturbance degree are ranked up the information to be searched, obtain corresponding related to the target search word
Degree and the maximum information to be searched of disturbance degree, the application is by Information Extraction Model and emotion model automatically to letter to be searched
The degree of correlation and disturbance degree of breath are ranked up, and can quickly, accurately and reliably obtain degree of correlation highest influences most important letter
There is wrong probability when ceasing, and a large amount of artificial investment can be saved, while artificial treatment can also be reduced.
In order to be extended by multi-C vector space to the initial ranging word that user inputs, to improve covering for search
Lid rate and accuracy are referring to fig. 2, also specific in an embodiment of the information processing method based on deep learning of the application
Include following content:
Step S201: it according to first position of the initial ranging word in multi-C vector space of user's input, obtains and institute
The distance for stating first position meets the expanded search word of default neighbor distance condition.
Step S202: the initial ranging word and the expanded search word are set as the target search word.
It is understood that position of the various words (i.e. target search word) in multi-C vector space is different, word
Meaning is closer, and the position in multi-C vector space is closer;On the contrary, if the meaning difference of word is very big, in multi-C vector
The position in space is remoter, and position vector of the initial ranging word inputted by user in multi-C vector space is found and its phase
Close (i.e. adjacent or close) term vector, these close words are set as expanded search word, with improve search coverage rate and
Accuracy.
In order to which expanded search word is supplemented and optimized by the entity relationship of knowledge mapping, to improve search
Coverage rate and accuracy, referring to Fig. 3, also have in an embodiment of the information processing method based on deep learning of the application
Body includes following content:
Step S301: by the initial ranging word inputted with user, corresponding first instance is expired in default knowledge mapping
The second instance of sufficient preset association relationship is set as supplementing expanded search word.
Step S302: vocabulary supplement is carried out to the expanded search word according to the supplement expanded search word and vocabulary is excellent
Change, obtains the expanded search word after vocabulary supplements and vocabulary optimizes.
It is understood that according to structural data, such as company, the equal information production disclosed in industrial and commercial organ of senior executive are known
The basic data for knowing map, the entity and relationship in knowledge mapping is extended by NER, and reduce in field article and news
Discrimination view, building obtain knowledge mapping, find the first instance in knowledge mapping by the initial ranging word that user inputs, and handle with
The second instance that first instance directly contacts is set as supplementing expanded search word, then can generate to by above-mentioned participle model
The expanded search word optimize and supplement.The expansion word of search term can be ranked up simultaneously, to improve search
Coverage rate and accuracy.
In order to be extended by multi-C vector space to the initial ranging word that user inputs, to improve covering for search
Lid rate and accuracy, in an embodiment of the information processing method based on deep learning of the application, it is also specific comprising just like
Lower content: position of each search term in multi-C vector space is determined according to default participle model.
Participle instruction is carried out according to the article of Baidupedia and target domain it is understood that Word2Vec can be used
Practice to which word is mapped to multi-C vector space, and determines that various words exist according to relationship of the word in sentence and article
Position in multi-C vector space;The meaning of word is closer, and the position in multi-C vector space is closer;On the contrary, if single
The meaning difference of word is very big, and the position in multi-C vector space is remoter.
In order to quickly, be accurately and reliably filtered to massive information, pluck choosing, understanding and matching treatment, this Shen
Please provide a kind of all or part of the content for realizing the information processing method based on deep learning based on depth
The embodiment of the information processing unit of habit, referring to fig. 4, the information processing unit based on deep learning specifically include as follows
Content:
Keyword abstraction module 10, for obtaining the keyword entity information of information to be searched according to Information Extraction Model.
Degree of correlation determining module 20, for being existed according to the keyword entity information and target search word of the information to be searched
Relative positional relationship in multi-C vector space determines the degree of correlation of the information to be searched Yu the target search word.
Disturbance degree determining module 30, for determining the information to be searched to the target search word according to emotion model
Disturbance degree.
Information determination module 40 to be searched, for according to the degree of correlation and the disturbance degree to the information to be searched into
Row sequence, obtains the degree of correlation corresponding with the target search word and the maximum information to be searched of disturbance degree.
As can be seen from the above description, the information processing unit provided by the embodiments of the present application based on deep learning, can pass through
Information Extraction Model obtains the keyword entity information of information to be searched, and is believed according to the keyword entity of the information to be searched
Breath and relative positional relationship of the target search word in multi-C vector space, determine the information to be searched and the target search
The degree of correlation of word, then according to emotion model determine the information to be searched to the disturbance degree of the target search word, further according to
The degree of correlation and the disturbance degree are ranked up the information to be searched, obtain corresponding related to the target search word
Degree and the maximum information to be searched of disturbance degree, the application is by Information Extraction Model and emotion model automatically to letter to be searched
The degree of correlation and disturbance degree of breath are ranked up, and can quickly, accurately and reliably obtain degree of correlation highest influences most important letter
There is wrong probability when ceasing, and a large amount of artificial investment can be saved, while artificial treatment can also be reduced.
In order to be extended by multi-C vector space to the initial ranging word that user inputs, to improve covering for search
Lid rate and accuracy are also specific referring to Fig. 5 in an embodiment of the information processing unit based on deep learning of the application
Include following content:
Expanded search word determination unit 51, initial ranging word for being inputted according to user in multi-C vector space
One position obtains the expanded search word for meeting default neighbor distance condition at a distance from the first position.
Target search word determination unit 52, for the initial ranging word and the expanded search word to be set as the mesh
Mark search term.
In order to which expanded search word is supplemented and optimized by the entity relationship of knowledge mapping, to improve search
Coverage rate and accuracy, referring to Fig. 6, also have in an embodiment of the information processing unit based on deep learning of the application
Body includes following content:
Expansion word determination unit 61 is supplemented, the initial ranging word for will input with user is in default knowledge mapping
The second instance that corresponding first instance meets preset association relationship is set as supplementing expanded search word.
Supplement optimization unit 62, for carrying out vocabulary supplement to the expanded search word according to the supplement expanded search word
Optimize with vocabulary, obtains the expanded search word after vocabulary supplements and vocabulary optimizes.
In order to be extended by multi-C vector space to the initial ranging word that user inputs, to improve covering for search
Lid rate and accuracy, in an embodiment of the information processing unit based on deep learning of the application, it is also specific comprising just like
Lower content: position map unit 71, for determining position of each search term in multi-C vector space according to default participle model
It sets.
In order to further explain this programme, the application also provides a kind of using the above-mentioned information processing based on deep learning
Device realizes the specific application example of the information processing method based on deep learning, specifically includes following content:
Step 1, information scratching grabs the mass data of internet and extracts the content of needs.
Step 2, participle model, using Word2Vec according to the text in field where Baidupedia and the information monitoring frame
The training of Zhang Jinhang participle is to be mapped to multi-C vector space for word.And it is true according to relationship of the word in sentence and article
Determine position of the various words in multi-C vector space.The meaning of word is closer, and the position in multi-C vector space is closer.
On the contrary, the position in multi-C vector space is remoter if the meaning difference of word is very big.
Step 3, knowledge mapping, according to structural data, such as company, the information that senior executive etc. discloses in industrial and commercial organ are constructed
Make the basic data of knowledge mapping.The entity and relationship in knowledge mapping are extended in field article and news by NER, and
And reduce discrimination view.The method comparative maturity of building knowledge mapping is not described in detail here.
Step 4, search term extends, by vector of the search term in multi-C vector space, find relative word to
Amount.Expansion word of the word for using these close as search term.
Step 5, search term sorts, and finds the entity in knowledge mapping, and the reality that entity is directly contacted by search term
Body is as related expanding word.The expansion word that can be generated in this way to participle model is optimized and is supplemented.It simultaneously can be to search
The expansion word of word is ranked up.
Step 6, information correlation model, by reading article, data engineering teacher summarizes the lists of keywords of article, thus
Obtain training data.With these training data training information extraction models.By Information Extraction Model, input article can be obtained
Lists of keywords.The positional relationship that multi-C vector space is carried out with these lists of keywords and search term and expansion word, can
To obtain the degree of correlation of search term Yu information to be searched.
Step 7, informational influence degree model obtains emotion mould as training data by the marking of the emotion of article and article
Type carries out emotion prediction to information with emotion model, obtains information to the disturbance degree of relevant search word.
It can be seen from the above, the application can also realize following advantageous effects:
(1) coverage for improving information analysis, improves the accuracy of information monitoring, by natural language processing technique and knows
Know map, automatically article can be analyzed.Analysis efficiency can achieve per hour a pieces up to a million (according to actual microcomputer
Cluster determines).It can be obtained more extensively by the search term extension in multi-C vector space and the entity relationship of knowledge mapping simultaneously
Search result.To reach the coverage rate that artificial treatment can not be compared.Secondly, by the word of natural language processing in text
The content of middle context carries out the article accuracy of analysis acquisition, and two generated when efficiently solving word composition word are anisotropic.
(2) important information is quickly found out by the sequence of information correlation and disturbance degree, passes through Information Extraction Model and feelings
Sense model is automatically ranked up the degree of correlation of information and disturbance degree.Data analyst more can quickly obtain the degree of correlation
Highest influences most important information list.
(3) cost of labor is reduced, since the process of information monitoring is automatically performed by microcomputer, a large amount of artificial throwing can be saved
Enter.The mistake occurred when can also be improved artificial treatment simultaneously.
Embodiments herein also provides the information processing method based on deep learning that can be realized in above-described embodiment
The specific embodiment of a kind of electronic equipment of middle Overall Steps, referring to Fig. 7, the electronic equipment specifically includes following content:
Processor (processor) 601, memory (memory) 602, communication interface (Communications
Interface) 603 and bus 604;
Wherein, the processor 601, memory 602, communication interface 603 complete mutual lead to by the bus 604
Letter;The communication interface 603 is for realizing the information processing unit based on deep learning, online operation system, client device
And other participate in the information transmission between mechanism;
The processor 601 is used to call the computer program in the memory 602, and the processor executes the meter
The Overall Steps in the information processing method based on deep learning in above-described embodiment are realized when calculation machine program, for example, described
Processor realizes following step when executing the computer program:
Step S101: the keyword entity information of information to be searched is obtained according to Information Extraction Model.
Step S102: according to the keyword entity information of the information to be searched and target search word in multi-C vector space
In relative positional relationship, determine the degree of correlation of the information to be searched Yu the target search word.
Step S103: determine the information to be searched to the disturbance degree of the target search word according to emotion model.
Step S104: being ranked up the information to be searched according to the degree of correlation and the disturbance degree, obtains and institute
State the corresponding degree of correlation of target search word and the maximum information to be searched of disturbance degree.
As can be seen from the above description, electronic equipment provided by the embodiments of the present application, can be obtained by Information Extraction Model to
The keyword entity information of information is searched for, and according to the keyword entity information and target search word of the information to be searched more
Relative positional relationship in dimensional vector space determines the degree of correlation of the information to be searched Yu the target search word, then root
The information to be searched is determined to the disturbance degree of the target search word, further according to the degree of correlation and the shadow according to emotion model
Loudness is ranked up the information to be searched, obtains the degree of correlation corresponding with the target search word and the maximum institute of disturbance degree
Information to be searched is stated, the application is by Information Extraction Model and emotion model automatically to the degree of correlation and disturbance degree of information to be searched
It is ranked up, can quickly, accurately and reliably obtain degree of correlation highest influences most important information, and can save a large amount of
Occurs the probability of mistake when manually putting into, while artificial treatment can also be reduced.
Embodiments herein also provides the information processing method based on deep learning that can be realized in above-described embodiment
A kind of computer readable storage medium of middle Overall Steps is stored with computer program on the computer readable storage medium,
The computer program realizes the whole of the information processing method based on deep learning in above-described embodiment when being executed by processor
Step, for example, the processor realizes following step when executing the computer program:
Step S101: the keyword entity information of information to be searched is obtained according to Information Extraction Model.
Step S102: according to the keyword entity information of the information to be searched and target search word in multi-C vector space
In relative positional relationship, determine the degree of correlation of the information to be searched Yu the target search word.
Step S103: determine the information to be searched to the disturbance degree of the target search word according to emotion model.
Step S104: being ranked up the information to be searched according to the degree of correlation and the disturbance degree, obtains and institute
State the corresponding degree of correlation of target search word and the maximum information to be searched of disturbance degree.
As can be seen from the above description, computer readable storage medium provided by the embodiments of the present application, can pass through information extraction
Model obtains the keyword entity information of information to be searched, and according to the keyword entity information and target of the information to be searched
Relative positional relationship of the search term in multi-C vector space determines that the information to be searched is related to the target search word
Degree, then determines the information to be searched to the disturbance degree of the target search word, further according to the correlation according to emotion model
Degree and the disturbance degree are ranked up the information to be searched, obtain the degree of correlation corresponding with the target search word and influence
The maximum information to be searched is spent, the application is by Information Extraction Model to emotion model automatically to the related of information to be searched
Degree and disturbance degree are ranked up, and can quickly, accurately and reliably obtain degree of correlation highest influences most important information, and can be with
There is the probability of mistake when saving a large amount of artificial investment, while artificial treatment can also be reduced.
All the embodiments in this specification are described in a progressive manner, same and similar portion between each embodiment
Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for hardware+
For program class embodiment, since it is substantially similar to the method embodiment, so being described relatively simple, related place is referring to side
The part of method embodiment illustrates.
It is above-mentioned that this specification specific embodiment is described.Other embodiments are in the scope of the appended claims
It is interior.In some cases, the movement recorded in detail in the claims or step can be come according to the sequence being different from embodiment
It executes and desired result still may be implemented.In addition, process depicted in the drawing not necessarily require show it is specific suitable
Sequence or consecutive order are just able to achieve desired result.In some embodiments, multitasking and parallel processing be also can
With or may be advantageous.
Although this application provides the method operating procedure as described in embodiment or flow chart, based on conventional or noninvasive
The labour for the property made may include more or less operating procedure.The step of enumerating in embodiment sequence is only numerous steps
One of execution sequence mode, does not represent and unique executes sequence.It, can when device or client production in practice executes
To execute or parallel execute (such as at parallel processor or multithreading according to embodiment or method shown in the drawings sequence
The environment of reason).
System, device, module or the unit that above-described embodiment illustrates can specifically realize by computer chip or entity,
Or it is realized by the product with certain function.It is a kind of typically to realize that equipment is computer.Specifically, computer for example may be used
Think personal computer, laptop computer, vehicle-mounted human-computer interaction device, cellular phone, camera phone, smart phone, individual
Digital assistants, media player, navigation equipment, electronic mail equipment, game console, tablet computer, wearable device or
The combination of any equipment in these equipment of person.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
In a typical configuration, calculating equipment includes one or more processors (CPU), input/output interface, net
Network interface and memory.
Memory may include the non-volatile memory in computer-readable medium, random access memory (RAM) and/or
The forms such as Nonvolatile memory, such as read-only memory (ROM) or flash memory (flash RAM).Memory is computer-readable medium
Example.
Computer-readable medium includes permanent and non-permanent, removable and non-removable media can be by any method
Or technology come realize information store.Information can be computer readable instructions, data structure, the module of program or other data.
The example of the storage medium of computer includes, but are not limited to phase change memory (PRAM), static random access memory (SRAM), moves
State random access memory (DRAM), other kinds of random access memory (RAM), read-only memory (ROM), electric erasable
Programmable read only memory (EEPROM), flash memory or other memory techniques, read-only disc read only memory (CD-ROM) (CD-ROM),
Digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape magnetic disk storage or other magnetic storage devices
Or any other non-transmission medium, can be used for storage can be accessed by a computing device information.As defined in this article, it calculates
Machine readable medium does not include temporary computer readable media (transitory media), such as the data-signal and carrier wave of modulation.
It will be understood by those skilled in the art that the embodiment of this specification can provide as the production of method, system or computer program
Product.Therefore, in terms of this specification embodiment can be used complete hardware embodiment, complete software embodiment or combine software and hardware
Embodiment form.
This specification embodiment can describe in the general context of computer-executable instructions executed by a computer,
Such as program module.Generally, program module includes routines performing specific tasks or implementing specific abstract data types, journey
Sequence, object, component, data structure etc..This specification embodiment can also be practiced in a distributed computing environment, in these points
Cloth calculates in environment, by executing task by the connected remote processing devices of communication network.In distributed computing ring
In border, program module can be located in the local and remote computer storage media including storage equipment.
All the embodiments in this specification are described in a progressive manner, same and similar portion between each embodiment
Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for system reality
For applying example, since it is substantially similar to the method embodiment, so being described relatively simple, related place is referring to embodiment of the method
Part explanation.In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ",
The description of " specific example " or " some examples " etc. means specific features described in conjunction with this embodiment or example, structure, material
Or feature is contained at least one embodiment or example of this specification embodiment.In the present specification, to above-mentioned term
Schematic representation be necessarily directed to identical embodiment or example.Moreover, description specific features, structure, material or
Person's feature may be combined in any suitable manner in any one or more of the embodiments or examples.In addition, in not conflicting feelings
Under condition, those skilled in the art by different embodiments or examples described in this specification and different embodiment or can show
The feature of example is combined.
The foregoing is merely the embodiments of this specification, are not limited to this specification embodiment.For ability
For field technique personnel, this specification embodiment can have various modifications and variations.It is all this specification embodiment spirit and
Any modification, equivalent replacement, improvement and so within principle should be included in the scope of the claims of this specification embodiment
Within.
Claims (10)
1. a kind of information processing method based on deep learning, which is characterized in that the described method includes:
The keyword entity information of information to be searched is obtained according to Information Extraction Model;
According to the keyword entity information of the information to be searched and relative position of the target search word in multi-C vector space
Relationship determines the degree of correlation of the information to be searched Yu the target search word;
Determine the information to be searched to the disturbance degree of the target search word according to emotion model;
The information to be searched is ranked up according to the degree of correlation and the disturbance degree, is obtained and the target search word pair
The maximum information to be searched of the degree of correlation and disturbance degree answered.
2. the information processing method according to claim 1 based on deep learning, which is characterized in that described according to
The keyword entity information of information to be searched and relative positional relationship of the target search word in multi-C vector space, determine described in
Before information to be searched and the degree of correlation of the target search word, comprising:
According to first position of the initial ranging word in multi-C vector space of user's input, obtain with the first position away from
From the expanded search word for meeting default neighbor distance condition;
The initial ranging word and the expanded search word are set as the target search word.
3. the information processing method according to claim 2 based on deep learning, which is characterized in that it is described will it is described just
Beginning search term and the expanded search word are set as before the target search word, comprising:
By the initial ranging word inputted with user, corresponding first instance meets default association pass in default knowledge mapping
The second instance of system is set as supplementing expanded search word;
Vocabulary supplement is carried out to the expanded search word according to the supplement expanded search word and vocabulary optimizes, is obtained by vocabulary
The expanded search word after supplement and vocabulary optimization.
4. the information processing method according to claim 2 based on deep learning, which is characterized in that described according to user
First position of the initial ranging word of input in multi-C vector space obtains meeting default phase at a distance from the first position
Before the expanded search word of neighborhood distance condition, comprising:
Position of each search term in multi-C vector space is determined according to default participle model.
5. a kind of information processing unit based on deep learning characterized by comprising
Keyword abstraction module, for obtaining the keyword entity information of information to be searched according to Information Extraction Model;
Degree of correlation determining module, for according to the keyword entity information of the information to be searched and target search word multidimensional to
Relative positional relationship in quantity space determines the degree of correlation of the information to be searched Yu the target search word;
Disturbance degree determining module, for determining influence of the information to be searched to the target search word according to emotion model
Degree;
Information determination module to be searched, for being arranged according to the degree of correlation and the disturbance degree the information to be searched
Sequence obtains the degree of correlation corresponding with the target search word and the maximum information to be searched of disturbance degree.
6. the information processing unit according to claim 5 based on deep learning, which is characterized in that further include:
Expanded search word determination unit, first in multi-C vector space of initial ranging word for being inputted according to user
It sets, obtains the expanded search word for meeting default neighbor distance condition at a distance from the first position;
Target search word determination unit, for the initial ranging word and the expanded search word to be set as the target search
Word.
7. the information processing unit according to claim 6 based on deep learning, which is characterized in that further include:
Expansion word determination unit is supplemented, the initial ranging word for will input with user is corresponding in default knowledge mapping
The second instance that first instance meets preset association relationship is set as supplementing expanded search word;
Supplement optimization unit, for carrying out vocabulary supplement and vocabulary to the expanded search word according to the supplement expanded search word
Optimization obtains the expanded search word after vocabulary supplements and vocabulary optimizes.
8. the information processing unit according to claim 6 based on deep learning, which is characterized in that further include:
Position map unit, for determining position of each search term in multi-C vector space according to default participle model.
9. a kind of electronic equipment including memory, processor and stores the calculating that can be run on a memory and on a processor
Machine program, which is characterized in that the processor realizes that Claims 1-4 is described in any item based on deep when executing described program
The step of spending the information processing method of study.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program quilt
The step of Claims 1-4 described in any item information processing methods based on deep learning are realized when processor executes.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910776373.0A CN110472018A (en) | 2019-08-22 | 2019-08-22 | Information processing method, device and computer storage medium based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910776373.0A CN110472018A (en) | 2019-08-22 | 2019-08-22 | Information processing method, device and computer storage medium based on deep learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110472018A true CN110472018A (en) | 2019-11-19 |
Family
ID=68512713
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910776373.0A Pending CN110472018A (en) | 2019-08-22 | 2019-08-22 | Information processing method, device and computer storage medium based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110472018A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111177405A (en) * | 2019-12-18 | 2020-05-19 | 深圳壹账通智能科技有限公司 | Data search matching method and device, computer equipment and storage medium |
CN113255349A (en) * | 2021-05-28 | 2021-08-13 | 北京字节跳动网络技术有限公司 | Information processing method and device and computer storage medium |
CN114088593A (en) * | 2021-10-11 | 2022-02-25 | 中科三清科技有限公司 | Method and device for determining sea salt discharge flux |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105404699A (en) * | 2015-12-29 | 2016-03-16 | 广州神马移动信息科技有限公司 | Method, device and server for searching articles of finance and economics |
CN106815252A (en) * | 2015-12-01 | 2017-06-09 | 阿里巴巴集团控股有限公司 | A kind of searching method and equipment |
CN107169010A (en) * | 2017-03-31 | 2017-09-15 | 北京奇艺世纪科技有限公司 | A kind of determination method and device of recommendation search keyword |
CN108470046A (en) * | 2018-03-07 | 2018-08-31 | 中国科学院自动化研究所 | Media event sort method and system based on media event search statement |
CN108932247A (en) * | 2017-05-24 | 2018-12-04 | 苏宁云商集团股份有限公司 | A kind of method and device optimizing text search |
CN109241238A (en) * | 2018-06-27 | 2019-01-18 | 广州优视网络科技有限公司 | Article search method, apparatus and electronic equipment |
-
2019
- 2019-08-22 CN CN201910776373.0A patent/CN110472018A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106815252A (en) * | 2015-12-01 | 2017-06-09 | 阿里巴巴集团控股有限公司 | A kind of searching method and equipment |
CN105404699A (en) * | 2015-12-29 | 2016-03-16 | 广州神马移动信息科技有限公司 | Method, device and server for searching articles of finance and economics |
CN107169010A (en) * | 2017-03-31 | 2017-09-15 | 北京奇艺世纪科技有限公司 | A kind of determination method and device of recommendation search keyword |
CN108932247A (en) * | 2017-05-24 | 2018-12-04 | 苏宁云商集团股份有限公司 | A kind of method and device optimizing text search |
CN108470046A (en) * | 2018-03-07 | 2018-08-31 | 中国科学院自动化研究所 | Media event sort method and system based on media event search statement |
CN109241238A (en) * | 2018-06-27 | 2019-01-18 | 广州优视网络科技有限公司 | Article search method, apparatus and electronic equipment |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111177405A (en) * | 2019-12-18 | 2020-05-19 | 深圳壹账通智能科技有限公司 | Data search matching method and device, computer equipment and storage medium |
CN113255349A (en) * | 2021-05-28 | 2021-08-13 | 北京字节跳动网络技术有限公司 | Information processing method and device and computer storage medium |
CN114088593A (en) * | 2021-10-11 | 2022-02-25 | 中科三清科技有限公司 | Method and device for determining sea salt discharge flux |
CN114088593B (en) * | 2021-10-11 | 2022-10-11 | 中科三清科技有限公司 | Method and device for determining sea salt discharge flux |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110298415B (en) | A kind of training method of semi-supervised learning, system and computer readable storage medium | |
CN108287864B (en) | Interest group dividing method, device, medium and computing equipment | |
Heymann et al. | Visual analysis of complex networks for business intelligence with gephi | |
US11372940B2 (en) | Embedding user categories using graphs for enhancing searches based on similarities | |
CN110472018A (en) | Information processing method, device and computer storage medium based on deep learning | |
US20170103337A1 (en) | System and method to discover meaningful paths from linked open data | |
Ngonmang et al. | Churn prediction in a real online social network using local community analysis | |
CN110232131A (en) | Intention material searching method and device based on intention label | |
Vishwakarma et al. | A comparative study of K-means and K-medoid clustering for social media text mining | |
Gao et al. | Chatbot or Chat-Blocker: Predicting chatbot popularity before deployment | |
CN115188067A (en) | Video behavior identification method and device, electronic equipment and storage medium | |
KR101910424B1 (en) | Method for movie ratings prediction using sentiment analysis of movie tags, recording medium and device for performing the method | |
CN112685573A (en) | Knowledge graph embedding training method and related device | |
Ilievski et al. | Does Wikidata Support Analogical Reasoning? | |
CN109582834B (en) | Data risk prediction method and device | |
CN110543635A (en) | information detection method and device based on deep learning and computer storage medium | |
Zhang et al. | Learning to build accurate service representations and visualization | |
Ye et al. | Applying the case difference heuristic to learn adaptations from deep network features | |
CN115660695A (en) | Customer service personnel label portrait construction method and device, electronic equipment and storage medium | |
Liu et al. | Automatic comic page image understanding based on edge segment analysis | |
CN110472140B (en) | Object word recommendation method and device and electronic equipment | |
CN111694995A (en) | Method and device for optimizing relational network and processing equipment | |
CN109977396A (en) | Emotion identification method, device, computer equipment and the computer storage medium of corpus participle | |
CN109948161A (en) | Data processing method and device for Chinese public sentiment | |
CN106157280B (en) | A kind of image processing method and device |
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 |