CN107480270A - A kind of real time individual based on user feedback data stream recommends method and system - Google Patents
A kind of real time individual based on user feedback data stream recommends method and system Download PDFInfo
- Publication number
- CN107480270A CN107480270A CN201710709794.2A CN201710709794A CN107480270A CN 107480270 A CN107480270 A CN 107480270A CN 201710709794 A CN201710709794 A CN 201710709794A CN 107480270 A CN107480270 A CN 107480270A
- Authority
- CN
- China
- Prior art keywords
- scene
- user
- storehouse
- data
- feature
- 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
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
By user feedback data stream, using the relational structure of hyperspace, realize that real time individual is recommended.Including three big core apparatus:User data device, feature database device, scene arrangement;The relation of the user data device and feature database device, user data device handling system data and search key data, scene analysis is realized by feature database device;The relation of the feature database device and scene arrangement, personalized identification is carried out by the hyperspace of feature database device, personalized recommendation is completed by scene arrangement.The affective characteristics storehouse, semantic feature storehouse and content characteristic storehouse are that self is inspired, and have memory and self-learning capability.Based on affective characteristics storehouse, semantic feature storehouse and content characteristic storehouse completion dimension step by step, feature and the decomposition of personalization and identification, accomplish what real time individual was recommended.
Description
Technical field
It is a kind of based on the real-time of user feedback data stream specifically the present invention relates to study personalized recommendation method
Personalized recommendation method and system.
Background technology
Under the social background of big data, so-called personalized recommendation is exactly the Characteristic of Interest and behavior according to user, Xiang Yong
Recommend information and content interested in family.Current, with the continuous expansion of information scale, content quantity and species quickly increase
There is so-called information overload problem, it is necessary to which the content for oneself wanting to buy can just be found by devoting a tremendous amount of time in length.In order to solve
This problem, personalized recommendation system arise at the historic moment.Personalized recommendation system finds of user by analyzing the behavior of user
Property demand, interest etc., then by user's information interested, Products Show to user.Commending system depends on data
Excavation and matching algorithm.
Search engine using Google, Baidu as representative can allow user to be exactly found oneself needs by inputting keyword
Relevant information.But if user can not accurate description oneself demand keyword, now search engine is with regard to helpless
.Different with search engine, commending system does not need user to provide clear and definite demand, but by analyzing the historical behavior of user
To be modeled to the interest of user, so as to actively recommend the information that can meet their interest and demand to user.Therefore, it is perfect
Commending system should be a kind of strong search engine.
Commending system has been widely used in many fields, wherein most typically and the neck with good development and application prospect
Domain is exactly e-commerce field, such as Jingdone district store, day cat store, Amazon shopping website.
Currently, most of personalized recommendations are that parsing and data cleansing are extracted in daily record using user data excavation, are trained,
With algorithm, personalized recommendation is realized.
Patent of the present invention, with reference to current domestic educational modernization present situation, the fundamental characteristics of user feedback data stream is analyzed, is carried
Go out a kind of mode of scene unit, displaying identification user behavior, realize personalized recommendation.
The content of the invention
It is an object of the invention to provide a kind of real time individual based on user feedback data stream to recommend method and system.This
The purpose of invention is to be achieved through the following technical solutions:
Patent of the present invention, by user feedback data stream, using the relational structure of hyperspace, realize that real time individual pushes away
Recommend.Including three big core apparatus:User data device, feature database device, scene arrangement;The user data device and feature database
The relation of device, user data device handling system data and search key data, scene point is realized by feature database device
Analysis;The relation of the feature database device and scene arrangement, personalized identification is carried out by the hyperspace of feature database device, by field
Scape device completes personalized recommendation.
The affective characteristics storehouse, semantic feature storehouse and content characteristic storehouse are that self is inspired, and have memory and self study energy
Power.
Based on affective characteristics storehouse, semantic feature storehouse and content characteristic storehouse completion dimension step by step, feature and personalization
Decompose and identify, accomplish what real time individual was recommended.
Further, scene unit:Using identifying layer by layer, standardized technique, certain behavior act is split as, such as:Place
It is reason, analysis, judgement, recommendation, alarm etc., each data behavior unit, pass through big data digging technology, place
Manage as a scene unit.
Further, content characteristic:Content characteristic is not scene unit, equivalent to the element of scene unit, scene unit
It is that the content characteristic included by it identifies, according to multidimensional model training, model checking, model prediction, extracts in effective
Hold feature;Scene is a kind of method analyzed and describe user's comprehensive characteristics, multi-dimensional spatial structure, directly portrays scene, is helped
User is helped to realize that personalization is portrayed;Content characteristic is the effective ways of abstract user behavior.
Further, real-time recommendation:Real-time calculating matching based on hyperspace model, not only including user's operand
According to the initiative of processing, in addition to user, search key and real-time computing capability, real-time scene change.Work as user
When scene changes, corresponding hyperspace, recommending data corresponding to acquisition are preferentially mapped to by new scene.
The present invention relates to the structure of system is as follows:
(001) customer traffic.Say in general sense, data flow is sequential, piecewise interval byte set, program
Receive or the file of disk and the read-write operation in the enterprising row data of network are arrived in storage, can be completed using data flow.
Patent of the present invention refers to the specific data block or memory cell of certain scene.For example, personalized once-through operation or a certain moment are used
Family personalized operation etc..It is summed up, is divided into two classes, system acquisition data and search key data.
(002) system acquisition data.Refer to the data flow of system interaction, strictly gathered in the way of system data flow, and
Stored by scene.Divided with reference to business function, such as:It is processing, analysis, judgement, recommendation, alarm etc.,
Each data storage cell, it is a scene unit, i.e., refines user behavior by service feature scene.Scene unit realization pair
The all standing of scene, the information of any recommendation results all possess the feature of scene unit completely.These features and service interaction rule
Model carries out standardized designs and storage one by one, can also be gathered by technological means, realization automatically updates.
(003) search key.Mainly the fractionation to Chinese key and word segmentation processing and carry out the management of adhering to separately property.Search
Keyword classification, basic source data is accumulated for follow-up semantic feature storehouse.Data content identification:Text class:Text, tape format
Text, number;Numeric class:Numerical value, the amount of money, percentage;It is compound:Keyword containing a variety of compositions of relations etc..Screen bar
Part and selective conditions identification, similar to the specification of forward direction expression, such as:Search courseware, video, operation, answer, explain, know
Know point etc..Integrated data content and screening conditions reduction concrete scene.
(004) sentiment analysis.By abstract emotion storehouse, analyze, quantify, complete emotion processing.Such as:Scene is positive and negative
Face:It is positive or passive;It is subjective or objective.This is just needed on the basis of sentiment analysis, excavates the category of learning state
Property and corresponding attribute emotion.The emotion of each scene is analyzed, forms affection data vocabulary of the learner to various contents, it is real
The self study of existing sentiment analysis.Emotion storehouse and sentiment analysis are closely bound up, bidirectional iteration circulation.
(005) semantic analysis.Current semantics analytical technology:First morphology and syntactic analysis, syntax and syntax tree are resettled, it is real
Existing semantics recognition, semanteme are determined by morphology and syntax.Semantic blind area be present in this mode:Morphology and syntax can not determine language completely
Justice.Patent of the present invention, using first semantic technology, by sentiment analysis and semantic comprehensive characteristics, word segmentation regulation corresponding to derivation, really
Determine clause, complete semantic understanding.Realize that displaying is cut.The final purpose of semantic analysis is to understand users ' individualized requirement.Sentence
Collection of the method structure for the feature of semantic tagger is most important.
(006) content is similar.Personalized core is classification, and the core of classification is that abstract, abstract result is content phase
Like property.Similitude in course collection of illustrative plates between all courses is connected using knowledge point, and similitude is single, lacks different reasons
Solution and extensibility rule.Similar source, which is excavated, two kinds, and one kind is user individual scene, recommends specific east according to characteristic scene
West;Also one kind is content similarities, recommends thing;Two kinds of combinations, realize various dimensions displaying similarity analysis, complete individual character
Change and recommend.Patent of the present invention, using hyperspace, from the different dimension such as sentiment analysis, semantic analysis, content knowledge point and
Displaying model, multi-angle find similitude, realize personalized recommendation.
(007) hyperspace model.Common space, refer to the three dimensions being made up of length.It is plus the time
Space-time.There are various hyperspace from the angle of science.Patent of the present invention refers to the hyperspace of capability model, specific bag
Include action trail, knowledge point, space attribute, time attribute, ability value, relevance etc..The hyperspace that these attributes sketch the contours,
Statistics is enumerated, attribute has more than six, at least sextuple space model.This scene that can recognize that with space characteristics, with multidimensional
Spatial model corresponds, and scene is unique, and spatial axes are supplemented or extended by the range of attributes in ability space, are allowed to reach
To the purpose of control scene unit personality.
(008) real-time recommendation.To be based on hyperspace model, in real time calculate matching, not only including user's operation data at
Reason, include the active request of user, search key and real-time computing capability, real-time scene change.When user's scene
When changing, corresponding hyperspace, recommending data corresponding to acquisition are preferentially mapped to by new scene.In strict accordance with pushing away
It is each user to take business, constantly, accurately pushes the service of personalization, while correct user's satisfaction based on feedback data stream
For degree until recommendation results reach or more increase than what is required, the lifting of quality is recommended in realization.This technological package, realization push away in real time
Recommend.
(009) complete.It is the end of a scene recommendation process.System acquisition user feedback data stream process, realize and use
User data automatically updates.
(010) affective characteristics storehouse.It is the semantic base that emotional ability is assigned on the basis of sentiment dictionary.As user is anti-
Feedback, teaching resource, user participate in exchange, interaction, and affective characteristics is one of important carrier for analyzing user behavior, and content is divided
Analysis, analysis of tendency etc. also just turn into emphasis.Patent application algorithm of the present invention enriches to existing sentiment dictionary, establishes a set of
Brand-new, that there is tendency degree, expansible sentiment dictionary and the affective characteristics storehouse of semantic relation.Affective characteristics storehouse is user
More careful, accurate tendentious personality analysis ability is provided, directly instructs personalized recommendation.
(011) semantic feature storehouse.As patent of the present invention is above-mentioned, by segmenting principle after first semantic, using Feature Words detection and
The participle technique of extraction.With reference to sentiment analysis and semantic feature storehouse, split and extract the dependence in content between critical behavior, behavior
Relation etc. hyperspace vector;The information such as inessential sentence or semanteme during identification is semantic.On this basis, feature is realized
Extraction and detection, finally realize semantic fitting according to vector relations.Experimental verification shows that the feature based on above method extraction is deposited
In interference content, finally by sentence member or semantic cleaning, the purification of semanteme is realized.This process possesses capability identification, checking
The flow such as scene hyperspace, fractionation, analysis, cleaning, fitting, the overall process realized the matching in semantic feature storehouse and found.
(011) content characteristic storehouse.Based on big data technology and the comprehensive understanding to user, abundant feature architecture, pass through
Dimension identification model is built, effectively differentiation user individual real-time requirement, the content characteristic of discoverable type are comprehensive and accurate to understand field
Scape is drawn a portrait, and realizes accurate personalized recommendation.Affective characteristics storehouse and the service function in semantic feature storehouse, all it is to surround a purpose:
User feedback formula displaying service.Content characteristic storehouse is exactly scene characteristic storehouse, builds complete displaying unit complete or collected works, uses field
Scape is a kind of method analyzed and describe user's comprehensive characteristics, multi-dimensional spatial structure, directly portrays scene, helps user to realize
Personalization is portrayed.
The beneficial effect that the present invention is brought:
1. model of place.Patent of the present invention is that self is inspired, and has memory, self-learning capability.Three big devices, number of users
According to device, scene arrangement, feature database device, dimension, feature and personalisation process are realized using scene hyperspace, is realized real-time
Personalized recommendation process.
2. scene unit.Patent utilization of the present invention identifies layer by layer, standardized technique, is split as certain behavior act, such as:
Processing, analysis, judging, recommending, alarm etc., each data behavior unit, by big data digging technology,
Complete scene unit identification.Scene unit greatly reinforces the ability of processing personalisation process.
3. content characteristic.Content characteristic is not scene unit, and equivalent to the element of scene unit, scene unit is to pass through it
Comprising content characteristic identification, according to multidimensional model training, model checking, model prediction, extract effective content characteristic.
Scape is a kind of method analyzed and describe user's comprehensive characteristics, multi-dimensional spatial structure, directly portrays scene, helps user to realize
Personalization is portrayed.Content characteristic is the effective ways of abstract user behavior.
4. real-time recommendation.That the real-time calculating based on hyperspace model matches, not only including user's operation data at
Reason, include the initiative of user, search key and real-time computing capability, real-time scene change.When user's scene
When changing, corresponding hyperspace, recommending data corresponding to acquisition are preferentially mapped to by new scene.
Brief description of the drawings
Fig. 1 is system structure diagram of the present invention.
Embodiment
The embodiment of the present invention is described in further detail with reference to Fig. 1.
As shown in figure 1, a kind of real time individual based on user feedback data stream recommends method and is designed by the present invention
System specifically comprises the following steps among actual application:
This step of step 001. realizes that two kinds of customer traffics merge, data acquisition and management work, into this step
Before rapid, system can gather user data by step 002 or step 003, and enter step 004;
Step 002. user feedback data stream has two kinds of data types, and this step completes collection and the place of system acquisition data
Reason, and enter step 001;
Step 003. user feedback data stream has two kinds of data types, and this step completes collection and the place of search key
Reason, and enter step 001;
Step 004. sentiment analysis, this step are identified by affective characteristics storehouse, complete sentiment analysis, and enter step 010
Or step 005;
Step 010. affective characteristics storehouse, this step identification affective characteristics simultaneously find feature, improve the process in affective characteristics storehouse,
And enter step 004;
Step 005. semantic analysis, this step are identified by semantic feature storehouse, complete semantic analysis, and enter step 011
Or step 006;
Step 011. semantic feature storehouse, this step analyze semantic feature and simultaneously find feature, improve the process in semantic feature storehouse,
And enter step 005;
Step 006. content is similar, and this step is identified by content characteristic storehouse, completes content analysis, and enter step 012
Or step 007;
Step 012. content characteristic storehouse, this step analysing content feature simultaneously find feature, improve the process in content characteristic storehouse,
And enter step 007;
Step 007. hyperspace model, this step are the hyperspace scene Recognition processes of capability model, complete scene
Analysis, and enter step 008;
Step 008. real-time recommendation, this step processing hyperspace model, calculates matching, recommends individualized content in real time
To the process of user, and enter step 009;
Step 009. is completed, and this step is the end of a scene recommendation process.System can enter user feedback data stream
Processing, the system automation renewal of user data is realized, the device or step 010 or step of arrow sensing of entering by flow circulation
011 or step 012;
Embodiments of the present invention are explained in detail above in conjunction with Fig. 1, but the present invention is not limited to above-mentioned implementation
Mode, can also be on the premise of present inventive concept not be departed from those of ordinary skill in the art's possessed knowledge
Make a variety of changes.
Claims (6)
1. a kind of real time individual based on user feedback data stream recommends method and system, it is characterised in that:Including three big cores
Center device:User data device, feature database device, scene arrangement;The relation of the user data device and feature database device, use
User data device handling system data and search key data, scene analysis is realized by feature database device;The feature database
The relation of device and scene arrangement, personalized identification is carried out by the hyperspace of feature database device, completed by scene arrangement individual
Propertyization is recommended.
2. a kind of real time individual based on user feedback data stream recommends method and system, it is characterised in that:The emotion is special
Sign storehouse, semantic feature storehouse and content characteristic storehouse are that self is inspired, and have memory and self-learning capability.
3. a kind of real time individual based on user feedback data stream recommends method and system, it is characterised in that:It is special based on emotion
Storehouse, semantic feature storehouse and content characteristic storehouse completions dimension step by step, feature and the decomposition of personalization and identification are levied, is accomplished real-time
Personalized recommendation.
4. a kind of real time individual based on user feedback data stream according to claim 1 recommends method and system, its
It is characterised by:Scene unit, using identifying layer by layer, standardized technique, certain behavior act is split as, such as:Processing, analysis
, judge, recommend, alarm etc., each data behavior unit, by big data digging technology, handle as one
Scene unit.
5. a kind of real time individual based on user feedback data stream according to claim 1 recommends method and system, its
It is characterised by:Content characteristic, content characteristic is not scene unit, and equivalent to the element of scene unit, scene unit is to pass through it
Comprising content characteristic identification, according to multidimensional model training, model checking, model prediction, extract effective content characteristic;
Scape is a kind of method analyzed and describe user's comprehensive characteristics, multi-dimensional spatial structure, directly portrays scene, helps user to realize
Personalization is portrayed;Content characteristic is the effective ways of abstract user behavior.
6. a kind of real time individual based on user feedback data stream according to claim 1 recommends method and system, its
It is characterised by:Real-time recommendation, be that the real-time calculating based on hyperspace model matches, not only including user's operation data at
Reason, include the initiative of user, search key and real-time computing capability, real-time scene change;When user's scene
When changing, corresponding hyperspace, recommending data corresponding to acquisition are preferentially mapped to by new scene.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710709794.2A CN107480270A (en) | 2017-08-18 | 2017-08-18 | A kind of real time individual based on user feedback data stream recommends method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710709794.2A CN107480270A (en) | 2017-08-18 | 2017-08-18 | A kind of real time individual based on user feedback data stream recommends method and system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107480270A true CN107480270A (en) | 2017-12-15 |
Family
ID=60601770
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710709794.2A Pending CN107480270A (en) | 2017-08-18 | 2017-08-18 | A kind of real time individual based on user feedback data stream recommends method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107480270A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109003478A (en) * | 2018-08-07 | 2018-12-14 | 广东小天才科技有限公司 | Learning interaction method and learning equipment |
CN109684466A (en) * | 2019-01-04 | 2019-04-26 | 钛氧(上海)教育科技有限公司 | A kind of intellectual education advisor system |
CN109994102A (en) * | 2019-04-16 | 2019-07-09 | 上海航动科技有限公司 | A kind of outer paging system of intelligence based on Emotion identification |
CN110120001A (en) * | 2019-05-08 | 2019-08-13 | 成都佳发安泰教育科技股份有限公司 | The method and system that a kind of knowledge based spectrum library mentions point in conjunction with memory curve |
-
2017
- 2017-08-18 CN CN201710709794.2A patent/CN107480270A/en active Pending
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109003478A (en) * | 2018-08-07 | 2018-12-14 | 广东小天才科技有限公司 | Learning interaction method and learning equipment |
CN109684466A (en) * | 2019-01-04 | 2019-04-26 | 钛氧(上海)教育科技有限公司 | A kind of intellectual education advisor system |
CN109684466B (en) * | 2019-01-04 | 2023-10-13 | 钛氧(上海)教育科技有限公司 | Intelligent education advisor system |
CN109994102A (en) * | 2019-04-16 | 2019-07-09 | 上海航动科技有限公司 | A kind of outer paging system of intelligence based on Emotion identification |
CN110120001A (en) * | 2019-05-08 | 2019-08-13 | 成都佳发安泰教育科技股份有限公司 | The method and system that a kind of knowledge based spectrum library mentions point in conjunction with memory curve |
CN110120001B (en) * | 2019-05-08 | 2021-07-20 | 成都佳发安泰教育科技股份有限公司 | Method and system for scoring based on combination of knowledge graph library and memory curve |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106874378B (en) | Method for constructing knowledge graph based on entity extraction and relation mining of rule model | |
CN111026671B (en) | Test case set construction method and test method based on test case set | |
CN110516067A (en) | Public sentiment monitoring method, system and storage medium based on topic detection | |
CN110347894A (en) | Knowledge mapping processing method, device, computer equipment and storage medium based on crawler | |
CN105630768B (en) | A kind of product name recognition method and device based on stacking condition random field | |
Ansari | Cost-based text understanding to improve maintenance knowledge intelligence in manufacturing enterprises | |
CN107480270A (en) | A kind of real time individual based on user feedback data stream recommends method and system | |
CN110516077A (en) | Knowledge mapping construction method and device towards enterprise's market conditions | |
CN110162632B (en) | Method for discovering news special events | |
CN111325029A (en) | Text similarity calculation method based on deep learning integration model | |
CN105843796A (en) | Microblog emotional tendency analysis method and device | |
CN105718585B (en) | Document and label word justice correlating method and its device | |
CN110019616A (en) | A kind of POI trend of the times state acquiring method and its equipment, storage medium, server | |
CN110442728A (en) | Sentiment dictionary construction method based on word2vec automobile product field | |
CN110134845A (en) | Project public sentiment monitoring method, device, computer equipment and storage medium | |
CN103177036A (en) | Method and system for label automatic extraction | |
CN117271767B (en) | Operation and maintenance knowledge base establishing method based on multiple intelligent agents | |
CN112966089A (en) | Problem processing method, device, equipment, medium and product based on knowledge base | |
CN109522396B (en) | Knowledge processing method and system for national defense science and technology field | |
CN110765277A (en) | Online equipment fault diagnosis platform of mobile terminal based on knowledge graph | |
CN110134844A (en) | Subdivision field public sentiment monitoring method, device, computer equipment and storage medium | |
Popping | Online tools for content analysis | |
Cui | Application of deep learning and target visual detection in English vocabulary online teaching | |
CN114840685A (en) | Emergency plan knowledge graph construction method | |
CN107527289B (en) | Investment portfolio industry configuration method, device, server and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20171215 |
|
WD01 | Invention patent application deemed withdrawn after publication |