CN110188208A - A kind of the information resources inquiry recommended method and system of knowledge based map - Google Patents
A kind of the information resources inquiry recommended method and system of knowledge based map Download PDFInfo
- Publication number
- CN110188208A CN110188208A CN201910481291.3A CN201910481291A CN110188208A CN 110188208 A CN110188208 A CN 110188208A CN 201910481291 A CN201910481291 A CN 201910481291A CN 110188208 A CN110188208 A CN 110188208A
- Authority
- CN
- China
- Prior art keywords
- user
- information resources
- interest
- resource
- entity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- 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
- G06F16/335—Filtering based on additional data, e.g. user or group profiles
- G06F16/337—Profile generation, learning or modification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- 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)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention proposes a kind of information resources of knowledge based map inquiry recommended methods and system, this method to pre-process first to knowledge mapping, knowledge mapping is mapped in the dense vector space of low-dimensional using expression learning method, the vector for obtaining entity indicates；Then user is calculated according to the historical behavior of user and user interest model is constructed to the interest-degree of information resources to the interest-degree of information resources, the vectorization expression of combining information resource and user；The accurate recommendation of information resources is realized by the similarity between computing resource and resource, user and resource.The present invention combines knowledge mapping expression study with user interest model to provide personalized service for user, take into account the inner link and user interest of knowledge, according to the resource name of user input query, recommend related to inquiry content to user and meet the information resources of user interest, so that more professional and specific aim is recommended in personalized inquiry.
Description
Technical field
The present invention relates to knowledge mapping and recommended technology fields, and in particular to a kind of information resources of knowledge based map are looked into
Ask recommended method and system.
Background technique
In recent years, the booming paces for having driven all trades and professions informationization of information technology, internet, Internet of Things, cloud
Calculating etc. gradually incorporates in daily life, and thus bring is the data of explosive growth.Huge information resources
Also the problem of resource overload is brought while library provides information abundant for user, this makes user in the interested letter of pick
It is taken considerable time in breath resource.And carry out personalized inquiry according to the historical behavior data of user and recommend, it can be effectively relieved
The problem of resource overload.
Recommender system is one of the effective means of current reply information overload, it analyzes user according to the historical behavior of user
Hobby, actively cater to his tastes, such as which kind of article user buys in various decision processes, read which news, which is listened first
Music.
Collaborative filtering proposes earliest, while being also a kind of research recommended technology most with application, it is relied on
In the behavior of user, pay close attention to being associated with for user and project, be broadly divided into two kinds of algorithms of different, be respectively algorithm based on user and
Project-based algorithm.Collaborative filtering basic principle based on user is exactly to find the user with similar behavior, is pushed away for user
Recommend the favorite resource of user institute having similar tastes and interests with it；Project-based collaborative filtering recommending is it is intended that user recommends with him once
Interested project has the project of similitude, similar to be not necessarily referring to the similar of the contents of a project, but is commented using user project
Valence or behavior, the similarity between excavation project.But collaborative filtering excessively relies on user behavior, causes to exist when system
New user or when new projects, recommendation will have no way of foundation.In addition to this, project has up to ten million kinds in real life, with user
The project for generating interaction often occupies the minority, and similar terms are only excavated to the behavior of project by user will lead to collaborative filtering calculation
The effect of method is poor.For this problem, the way of current most of researchs is to introduce auxiliary information as the defeated of proposed algorithm
Enter.
And knowledge mapping contains semantic information abundant, it is intended to indicate the reality in real world in the form of structuring
Body or concept and the incidence relation between them, essence are a huge semantic network figures, by mass knowledge with more straight
The mode of sight is shown in front of the user, is made of node and side, wherein node on behalf entity or concept, between Bian Daibiao entity
Relationship or entity attributes.Knowledge mapping introduces more semantic relations, provides different relationship connection types, will know
Know map to be introduced into recommender system, semantic information abundant in the map that can turn one's knowledge to advantage, so as to find profoundly
User interest avoids recommendation results from being confined to single type, improves recommender system accuracy, diversity and interpretation, from
And user is improved to the satisfaction of recommendation results.
Have the research of the recommended method of some knowledge based maps, such as the recommended method based on path at present, needs
The method for constructing a specific path of two entities of connection, but constructing path manually is difficult to reach optimal in practice；Base
In the recommended method of nomography is intuitively semantic network figure using knowledge mapping the characteristics of, using random walk scheduling algorithm in figure
Node is sampled, but portable poor, the computation complexity height of nomography, when facing large-scale knowledge mapping, is difficult to accomplish reality
When calculate.
Summary of the invention
Goal of the invention: in view of the drawbacks of the prior art and insufficient, the present invention provides a kind of information money of knowledge based map
Recommended method is inquired in source, takes into account the inner link and user interest of knowledge, quickly high according to the resource name of user input query
Recommend related to inquiry content to user and meet the information resources of user interest in effect ground.
Technical solution: according to the first aspect of the invention, the information resources inquiry for providing a kind of knowledge based map is recommended
Method the described method comprises the following steps:
(1) it indicates that learning method maps to knowledge mapping in the dense vector space of low-dimensional using knowledge mapping, realizes
To the vectorization semantic expressiveness of the information resources in knowledge mapping；
(2) according to user's history behavior, user is calculated to the interest-degree of information resources；
(3) it combines user to the interest-degree of information resources and the vectorization semantic expressiveness of information resources, constructs user interest
Model；
(4) according to the information resources of user query, the similarity of the information resources Yu other information resource is calculated, is taken similar
The information resources for spending TOP-M form candidate resource collection；
(5) similarity for calculating information resources and user that candidate resource is concentrated filters out similar from candidate resource concentration
The information resources for spending TOP-N form recommendation list.
Further, the step 1 includes:
(11) triple (h, r, t) of specified quantity, referred to as positive example triple are chosen from knowledge mapping, wherein h, t
An entity, tail entity are respectively represented, r indicates the relationship between two entities；
(12) using the head entity or tail entity of negative sampling algorithm replacement positive example triple, negative example triple is obtained；
(13) using learning model repetitive exercise positive example triple and negative example triple is indicated to restraining, obtain entity to
Amount indicates Vi={ v1,v2……,vm, wherein m indicates dimension.
Further, the step 12 includes:
(121) in all triples of relationship r, the mean number of the corresponding tail entity of every head entity is counted, is denoted as
tph；The mean number for counting the corresponding head entity of each tail entity, is denoted as hpt；
(122) for a positive example triple (h, r, t), entity is extracted to replace an entity h and tail entity t, with the general of p
Rate replaces head entity, replaces tail entity with the probability of 1-p, generates negative example triple, wherein the calculation formula of replacement Probability p are as follows:
Further, the step 2 includes:
(21) collect include user behavior log, including user browsing resource name, resource content length, browsing when
It is long；
(22) according to whether click browsing, browsing time, surfing establish multiple linear equation, user is to resource for calculating
Interest-degree.
Further, the step 22 includes:
(221) user clicks certain information resources i of browsing, and remembering that it clicks interest-degree is Ci；
(222) according to user to the browsing duration t of resource iiIts navigation interest degree is calculated with the average surfing of user
Ri:
Wherein t1Indicate minimum browsing time of the user to resource i, t2Indicate maximum browsing time of the user to resource i, S
For the average surfing of user,L is the total length that user browses resource, and T is the total time that user browses resource；
(223) comprehensive to click interest-degree and navigation interest degree, user is obtained to the interest-degree I of resource ii=ω1Ci+ω2Ri,
Wherein ω1、ω2It represents and clicks interest-degree and the navigation interest degree weight shared when calculating total interest-degree, and ω1+ω2=1.
Further, user interest model in the step 3 are as follows:WhereinWeight shared by the past interest vector of user is represented,Represent weight shared by current interest vector, UpresentIt indicates
Currently updated interest vector indicates user, UpreviousIndicate that user goes over the vector expression of interest, IiIndicate user to i-th
The interest-degree of resource, ViIndicate that the vector of i-th resource indicates.
Further, the method is after step 1 further include: according to the distance between the vector computing resource of information resources,
According to its similarity of Distance Judgment, similar information resources entity aggregates are formed into a cluster, different information resources entity is drawn
It assigns in different clusters.
Further, the similarity between two resources is calculated by COS distance in the step 4, in the step 5
The similarity between information resources and user interest is calculated by COS distance.
According to the second aspect of the invention, a kind of information resources inquiry recommender system of knowledge based map is provided, it is described
System includes:
Data preprocessing module, for indicating that learning model is empty by knowledge mapping insertion low-dimensional vector using knowledge mapping
Between, it is indicated by the vectorization that study obtains entity, relationship and attribute；
User interest model constructs module, for analyzing user behavior, understands the interest of user, building user is emerging
Interesting model；And
Recommending module is inquired, for the resource acquisition candidate resource collection according to user input query, is concentrated in candidate resource
The resource for filtering out interest of being close to the users is recommended.
The utility model has the advantages that the present invention is based on knowledge mappings to indicate that the recommended method of study is rich as a language using knowledge mapping
Data set rich, that logical reasoning ability is strong is dissolved into traditional proposed algorithm, is learnt using expression by each of knowledge mapping
Entity and relationship are expressed as dense low-dimensional real-valued vectors, reduce the higher-dimension of knowledge mapping, so that in low-dimensional vector space, it can
It with the semantic relation between efficient computational entity, reduces due to introducing knowledge mapping bring extra computation burden, so that enhancing is known
Know the flexibility of application of the graphic chart.It is embodied in:
1, semantic information abundant in knowledge mapping is taken full advantage of, traditional collaborative filtering is made up and does not consider to be recommended
The defect of the semantic information of project.Indicate that study is dense to low-dimensional the entity in knowledge base, relationship map using knowledge mapping
Vector space in, complete that the semantic expressiveness of entity and relationship has been obviously improved computational efficiency, can pass through COS distance degree
The semantic similarity between entity is measured, while an entity has a dense vector corresponding, also alleviates Sparse
Problem.
2, during model training, using the negative sampling algorithm of Bernoulli Jacob, the algorithm is real by the way that different replacement heads is arranged
The probability of body or tail entity effectively avoids introducing the negative example triple of mistake.
3, user interest model is constructed on the basis of analyzing user's history behavior, user interest is mapped into low-dimensional vector
In space, so that user interest and information resources become the point in vector space, pass through computing resource and resource, user and resource
The distance between known to its similarity, calculating process is concise.
4, knowledge mapping expression study is combined with user interest model to provide personalized service for user, takes into account and knows
The inner link and user interest of knowledge are recommended related simultaneously to inquiry content according to the resource name of user input query to user
And meet the information resources of user interest, so that more professional and specific aim is recommended in personalized inquiry.
Detailed description of the invention
Fig. 1 is the recommended method overall flow figure according to the embodiment of the present invention；
Fig. 2 is the log integrity result figure according to the embodiment of the present invention；
Fig. 3 user interest model according to an embodiment of the present invention constructs flow chart；
Fig. 4 is the recommender system module map according to the embodiment of the present invention.
Specific embodiment
Technical solution of the present invention is described further with reference to the accompanying drawing.It is to be appreciated that examples provided below
Merely at large and fully disclose the present invention, and sufficiently convey to person of ordinary skill in the field of the invention
Technical concept, the present invention can also be implemented with many different forms, and be not limited to the embodiment described herein.For
The term in illustrative embodiments being illustrated in the accompanying drawings not is limitation of the invention.
In one embodiment, by taking Water Conservancy Information Resources inquiry is recommended as an example, water conservancy neck is introduced in inquiry recommendation process
Domain knowledge map indicates that learning method arrives the entity in knowledge mapping, relationship map as auxiliary information, using knowledge mapping
In the dense vector space of low-dimensional, realizes to the semantic expressiveness of entity and relationship, make up conventional recommendation algorithm and do not consider semantic letter
The defect of breath.Then, on the basis of analyzing user browsing behavior and browsing content, the user interest model of low-dimensional is constructed.Most
Afterwards, knowledge mapping and user interest model are combined, constructs the information resources based on Water Resources Domain knowledge mapping and inquire and recommends
System is realized and precisely recommends the Water Conservancy Information Resources for meeting user interest according to the inquiry of user.
Fig. 1 is the flow chart that the information resources based on Water Resources Domain knowledge mapping inquire recommended method, as shown in Figure 1, should
The realization process of method the following steps are included:
Step 1, indicate that learning model maps to knowledge mapping in the dense vector space of low-dimensional using knowledge mapping, it is real
Now to the vectorization semantic expressiveness of the Water Conservancy Information Resources in knowledge mapping.
Knowledge mapping is using information resources as its conceptual entity node, using the correlation attribute information of information resources as its spy
Label node is levied, the side of two nodes represents relationship or entity attributes between entity.Using triple indicate be (h,
R, t) or (e, a, v), wherein h, t in (h, r, t) respectively represent an entity, tail entity, r indicates the relationship between two entities,
E in (e, a, v) represents entity, and a, v represent entity attributes and attribute value.
Specifically, step 1 includes:
Step 1.1, a certain number of triples (h, r, t), referred to as positive example triple are chosen from knowledge mapping, such as
Triple (Yantan reservoir, project scale, big 1 type) in Water Resources Domain knowledge mapping；
Step 1.2, using the head entity or tail entity of negative sampling algorithm replacement positive example triple, negative example ternary is generated
Group, i.e. wrong positive example triple, specific steps are as follows:
Step 1.2.1 counts the mean number of the corresponding tail entity of every head entity in all triples of relationship r,
It is denoted as tph；The mean number for counting the corresponding head entity of each tail entity, is denoted as hpt；
For example, for there are 3 triple (h1,r,t1), (h1,r,t2), (h2,r,t3) knowledge mapping, h1It is corresponding
Tail entity number is 2, h2Corresponding tail entity number is 1, thenSimilarly,
Step 1.2.2, definition replacement new probability formulaIt can be obtained by above-mentioned tph and hpt
Step 1.2.3 extracts the entity in knowledge mapping for a positive example triple (h, r, t) to replace positive example three
Head entity h or tail entity t in tuple, to generate a new triple, this triple is considered as negative example ternary
Group.Head entity is replaced with the probability of p, negative example ternary is generated to break positive example triple with the probability replacement tail entity of 1-p
Group.
Step 1.3, it is extremely restrained using expression learning model repetitive exercise positive example triple and negative example triple.
Indicate learning model by continuous iteration come the parameter in more new model, i.e. loss function, after successive ignition
Loss convergence acquires the minimum value of loss function, entity vector, relation vector and attribute vector loop convergence to optimal, damage
Function is lost to be defined as follows:
Wherein (h, r, t) indicates positive example triple, and (h ', r, t ') indicates negative example triple, and γ is the marginal value of setting, symbol
Number []+It is hinge loss function.‖ h+r-t ‖ is the sum of vector for indicating head entity h and relationship r and the difference of tail entity t vector
Distance, it is expected that the value of ‖ h+r-t ‖ is the smaller the better, then it is expected ‖ for negative example triple in a model for a positive example triple
The value of h '+r-t ' ‖ is the bigger the better, and can distinguish positive negative sample by training pattern in an experiment in this way.
In order to accelerate to restrain, initialization and normalized are carried out to data.The vector of entity, relationship is carried out first equal
Even distribution initialization:
K indicates specified vector dimension, after initialization, is normalized:
In iterative process each time, require that first entity vector is normalized.During model training
If it is very big that all triples are all iterated with training so cost, therefore in order to accelerate to restrain, using small lot
Training algorithm of the gradient descent algorithm as model is chosen small quantities of that is, in iteration each time from Water Resources Domain knowledge mapping
The training triple of amount is used as with reference to being trained, then constant by a learning rate to determine the more new direction of model
Gradient steps come undated parameter, i.e. loss function.
For model training to when restraining, the entity in knowledge mapping can be mapped to the corresponding position in lower dimensional space, so that
Physical distance with same alike result or identical relationship is close, and model training terminates.The vector for obtaining entity indicates Vi={ v1,
v2……,vk}。
Step 2, information resources are clustered.
According to the vector of information resources, COS distance between computing resource judges its similarity, by similar information resources
Entity aggregates get up to be formed a cluster, and different information resources entity is divided into different clusters, similar information resources
That is the information resources with same alike result or identical relationship, and different information resources are not present or there are less same alike results
Or the information resources of identical relationship.After cluster, similar information resources entity is polymerized to a cluster, is candidate in subsequent step
The screening of resource set reduces calculation amount, effectively improves efficiency.
Step 3, according to user's history behavior, user is calculated to the interest-degree of information resources, the specific steps are as follows:
Step 3.1, collect include user behavior log, it is resource name including user's browsing, resource content length, clear
Look at the information such as duration；
Usage log component carries out log recording in embodiment, buries a technology using js and collects user's browsing water conservancy information money
The behavioral data in source.The operating condition of system is contained in log, such as the connection state of database, the error message of system, such as
The customized log of mistake and user inside server or program exports content, such as the Debugging message of program, user's
Behavioral data etc..According to demand, it needs to be filtered system log processing, obtains log only comprising user behavior data.
Fig. 2 is to pass through pretreated log, wherein _ ip is the address ip of user, this is the unique identification of user；_
Url is the current address URL；_ refer is the address URL of the upper access of user；_ millisecond is that user enters the page
Millisecond number, this be long type the millisecond number since 1970.1.1, facilitate calculating time interval；_ id is browsed by user
Water Conservancy Information Resources unique identification；_ name is the title of Water Conservancy Information Resources；_ length is that the content of information resources is long
Degree, what is indicated is the length of information resources summary info.
Step 3.2, from clicking browsing, browsing three duration, surfing aspects, using multiple linear equation as base
User interest degree is abstracted into number by plinth, calculates user to the interest-degree of resource.Specific steps are as follows:
Step 3.2.1, when user clicks certain information resources i of browsing, remembering that it clicks interest-degree is Ci；
User, only there are two types of operation, clicks and does not click on, by the click interest-degree of user to resource is defined as:
Browsing " Danjiangkou power station " this Water Conservancy Information Resources are clicked from user known to Fig. 2 log, remember Ci=1.
Step 3.2.2 calculates its navigation interest degree R according to user's browsing time, surfingi, wherein surfing can
According to browsing duration and resource content length computation；
The time that user browses each resource is longer, shows that user is higher to the interest-degree of the resource；Otherwise the time is shorter,
Illustrate that the user is lower to the interest-degree of the resource.And in the case where the average speed that user reads substantially is stablized, user
The speed for browsing each resource is slower, illustrates that it spends more time to go to read, can determine whether that it is higher to the interest-degree of the resource；
Otherwise its interest-degree is lower.By the navigation interest degree of user is defined as:
Wherein tiIndicate that user browses the duration of resource i, t1It indicates the minimum browsing time, works as ti<t1When, indicate user couple
The resource it is not interested or it is overdue enter resource details page；t2It indicates the maximum browsing time, works as ti>t2When, it is believed that user is possible to
It is to rest on the page in navigation process and go to handle other things.In order to avoid these situations influence the meter of user interest degree
It calculates, navigation interest degree the case where these is calculated as 0.S is the average surfing of user, not due to different user's reading abilities
Together, but its reading rate is stable, the historical behavior of foundation user, and the average browsing speed of user is calculated according to the following formula
Degree:
Wherein L is the total length that user browses resource, and T is the total time that user browses resource.
It is 119 words from the content-length " _ length " of the resource known to Fig. 2 log, and the browsing time can be according to front and back two
The time difference of secondary behavior subtracts each other to obtain, and browsing resource time t is about 35 seconds.For example, it is assumed that user is averaged, surfing S is 7.5
Navigation interest degree R is calculated in word/secondiIt is 2.21.
Step 3.2.3, comprehensive click interest-degree and navigation interest degree, user are defined as I to the interest-degree of resource ii=ω1Ci+ω2Ri, wherein ω1、ω2It represents and clicks interest-degree and the navigation interest degree weight shared when calculating total interest-degree, and ω1
+ω2=1.
For example, ω1、ω2Value 0.2,0.8 respectively, finally obtains user to the synthesis interest-degree I of the resourceiIt is 1.97.
Step 4, it is emerging that user is constructed to the interest-degree of information resources and the vectorization semantic expressiveness of information resources in conjunction with user
Interesting model.
User interest is expressed as dense low-dimensional real-valued vectors, dimension and information resources entity vector by user interest model
Identical, purpose is exactly in the lower dimensional space where user interest to be mapped to entity, and defined formula is U={ u1,u2……,um}。
Fig. 3 is that user interest model constructs process, in conjunction with user to the interest-degree I of information resources iiWith information resources i's
Vectorization semantic expressiveness Vi={ v1,v2……,vm, user interest model is constructed, formula is, WhereinWeight shared by the past interest vector of user is represented,Represent current interest vector
Shared weight, UpresentIndicate that currently updated interest vector indicates user, UpreviousIndicate user go over interest to
Amount expression, IiIndicate interest-degree of the user to i-th resource, ViIndicate that the vector of i-th resource indicates.
Finally, knowledge mapping and user interest are mapped in lower dimensional space, so that user interest and Water Conservancy Information Resources
Entity becomes the point in lower dimensional space, can judge its similarity by the distance between computing resource and resource, resource and user.
Step 5, according to the information resources of user query, the similarity of the information resources Yu other information resource is calculated, is taken
The information resources of similarity TOP-M form candidate resource collection.
Candidate resource collection refers to the set of analog information resource, is the source of the information resources in consequently recommended list, it
It ensure that consequently recommended information resources are related to user query content.In low-dimensional vector space, COS distance meter can be passed through
Calculate the similarity between two resources, calculation formula are as follows:
Wherein etIt is in knowledge mapping in addition to entity eiOther entities in addition, Vi、VtRepresent entity eiWith entity etTo
Amount.It is obtained and Water Conservancy Information Resources entity e by calculatingiThe M Water Conservancy Information Resources with higher similarity form candidate money
Source collection D={ d1,d2,……,dM}。
Step 6, the similarity for calculating information resources and user that candidate resource is concentrated filters out phase from candidate resource concentration
Recommendation list is formed like the information resources of degree TOP-N.
Candidate resource collection is the premise for generating consequently recommended list, which ensure that recommendation is related to inquiry content, and
There is provided personalized ventilation system, it is also necessary to filter out interest of being close to the users from candidate resource concentration in conjunction with user interest model
Resource generates consequently recommended list.
In same lower dimensional space, if user and resource is closely located, then it represents that favorable rating of the user to the resource
It is high；Conversely, illustrating that user is not concerned with the resource if user is apart from each other with resource.User is calculated using cosine similarity formula
The similarity for the Water Conservancy Information Resources concentrated with candidate resource, filtering out from candidate resource concentration has higher phase with user interest
Like N number of entity of degree, generates TOP-N and recommend.
Fig. 4 is that the information resources of knowledge based map inquire the module map of recommender system, including data preprocessing module, use
Family interest model constructs module, inquiry recommending module, and the data preprocessing module includes that knowledge mapping indicates unit, using knowing
Knowing map indicates that knowledge mapping is embedded in low-dimensional vector space by learning model, by study obtain entity, relationship and attribute to
Quantization means；The user interest model building module is analyzed according to user behavior, is understood the interest of user, is constructed user
Interest model；The inquiry recommending module is mainly the resource according to user input query, to obtain candidate resource collection, in candidate
The resource for interest of being close to the users is filtered out in resource set.
Specifically, knowledge mapping indicates that unit chooses a certain number of positive example triples (h, r, t) from knowledge mapping,
Using the head entity or tail entity of negative sampling algorithm replacement positive example triple, negative example triple is generated, is instructed using model iteration
Practice positive example triple and negative example triple to restraining, realizes the corresponding positions mapped to the entity in knowledge mapping in lower dimensional space
It sets, the vector for obtaining entity indicates Vi={ v1,v2……,vk}。
As preferred embodiment, data preprocessing module further includes cluster cell, is provided using clustering method to information
Source entity is clustered.Cluster cell is according to the vectors of information resources, and COS distance between computing resource judges its similarity,
Similar information resources entity aggregates are got up to be formed a cluster, and different information resources entity is divided into different clusters
In, similar information resources are to have the information resources of same alike result or identical relationship, and different information resources are not present
Or there are less same alike result or the information resources of identical relationship.
It includes log collection unit, journal processing unit, log analysis unit, Yong Huxing that user interest model, which constructs module,
Interesting model construction unit, wherein log of the log collection unit collection comprising user behavior, the resource name browsed including user,
The information such as resource content length, browsing duration；Journal processing unit is filtered processing to system log, obtains only comprising user
The log of behavioral data；Log analysis unit is from clicking browsing, browsing three duration, surfing aspects, with polynary line
Property equation based on, user interest degree is abstracted into number, calculates user to the interest-degree of resource；User interest model building is single
Member combines user to the interest-degree of information resources and the vectorization semantic expressiveness of information resources, constructs user interest model.
Inquiring recommending module includes candidate resource collection unit, recommendation list unit, and wherein candidate resource collection unit passes through remaining
Chordal distance calculates the similarity between two resources, and the information resources of similarity TOP-M is taken to form candidate resource collection；Recommendation list
Unit calculates the similarity of the information resources of user and candidate resource concentration using cosine similarity formula, concentrates from candidate resource
N number of entity that there is higher similarity with user interest is filtered out, TOP-N is generated and recommends.
Specific formula for calculation involved in above-mentioned each module can refer to respective formula in embodiment of the method, no longer superfluous herein
It states.
The present invention provides a kind of information resources of knowledge based map inquiry recommended method and recommender systems, by knowledge graph
Spectral representation study is combined with user interest model, is realized and is precisely recommended the information for meeting user interest to provide according to the inquiry of user
Source.Knowledge mapping is semantic data set abundant, logical capability is strong, contains structuring and unstructured data, gos deep into knowledge
Inner link.Inclusion knowledge and user preference are fused together by the present invention, so that proposed algorithm is more authoritative, professional
And specific aim.Also, the recommender system for having merged knowledge mapping can also provide explanation, and user or system designer is allowed to know
Why recommend these projects, helps to improve the user satisfaction of efficiency, convincingness and recommender system.Indicate that study refers to
Using number, such as matrix, vector, to express certain things of real world, this expression way be conducive to subsequent classification or
Decision problem allows follow-up work to get twice the result with half the effort.Similarly, knowledge mapping indicates that study is intended to be by entity and transformation
Vector in lower dimensional space, while not changing the immanent structure of knowledge mapping.It is obviously improved by entity and relation vector
Computational efficiency can measure the semantic similarity between entity by Euclidean distance or COS distance, while an entity has one
The problem of a dense vector is corresponding, also alleviates Sparse.
The above is only a preferred embodiment of the present invention, it should be pointed out that: for the ordinary skill people of the art
For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also answered
It is considered as protection scope of the present invention.
Claims (10)
1. a kind of information resources of knowledge based map inquire recommended method, which is characterized in that the described method comprises the following steps:
(1) it indicates that learning method maps to knowledge mapping in the dense vector space of low-dimensional using knowledge mapping, realizes to knowing
Know the vectorization semantic expressiveness of the information resources in map；
(2) according to user's history behavior, user is calculated to the interest-degree of information resources；
(3) it combines user to the interest-degree of information resources and the vectorization semantic expressiveness of information resources, constructs user interest model；
(4) according to the information resources of user query, the similarity of the information resources Yu other information resource is calculated, similarity is taken
The information resources of TOP-M form candidate resource collection；
(5) similarity for calculating information resources and user that candidate resource is concentrated filters out similarity from candidate resource concentration
The information resources of TOP-N form recommendation list.
2. the information resources of knowledge based map according to claim 1 inquire recommended method, which is characterized in that the step
Rapid 1 includes:
(11) triple (h, r, t) of specified quantity, referred to as positive example triple are chosen from knowledge mapping, wherein h, t distinguish
Head entity, tail entity are represented, r indicates the relationship between two entities；
(12) using the head entity or tail entity of negative sampling algorithm replacement positive example triple, negative example triple is obtained；
(13) using the extremely convergence of learning model repetitive exercise positive example triple and negative example triple is indicated, the vector table of entity is obtained
Show Vi={ v1, v2..., vm, wherein m indicates dimension.
3. the information resources of knowledge based map according to claim 2 inquire recommended method, which is characterized in that the step
Rapid 12 include:
(121) in all triples of relationship r, the mean number of the corresponding tail entity of every head entity is counted, tph is denoted as；
The mean number for counting the corresponding head entity of each tail entity, is denoted as hpt；
(122) for a positive example triple (h, r, t), entity is extracted to replace an entity h and tail entity t, is replaced with the probability of p
An entity is changed, tail entity is replaced with the probability of 1-p, generates negative example triple, wherein the calculation formula of replacement Probability p are as follows:
4. the information resources of knowledge based map according to claim 2 inquire recommended method, which is characterized in that the step
Rapid 2 include:
(21) log comprising user behavior is collected, resource name, resource content length, browsing duration including user's browsing；
(22) according to whether click browsing, browsing time, surfing establish multiple linear equation, user is to the emerging of resource for calculating
Interesting degree.
5. the information resources of knowledge based map according to claim 4 inquire recommended method, which is characterized in that the step
Rapid 22 include:
(221) user clicks certain information resources i of browsing, and remembering that it clicks interest-degree is Ci；
(222) according to user to the browsing duration t of resource iiIts navigation interest degree R is calculated with the average surfing of useri:
Wherein t1Indicate minimum browsing time of the user to resource i, t2User is indicated to the maximum browsing time of resource i, S is to use
The average surfing at family,L is the total length that user browses resource, and T is the total time that user browses resource；
(223) comprehensive to click interest-degree and navigation interest degree, user is obtained to the interest-degree I of resource ii=ω1Ci+ω2Ri, wherein
ω1、ω2It represents and clicks interest-degree and the navigation interest degree weight shared when calculating total interest-degree, and ω1+ω2=1.
6. the information resources of knowledge based map according to claim 5 inquire recommended method, which is characterized in that the step
User interest model in rapid 3 are as follows:WhereinRepresent the past interest vector of user
Shared weight,Represent weight shared by current interest vector, UpresentIndicate user's currently updated interest vector
It indicates, UpreviousIndicate that user goes over the vector expression of interest, IiIndicate interest-degree of the user to i-th resource, ViIndicate i-th
The vector of resource indicates.
7. the information resources of knowledge based map according to claim 1 inquire recommended method, which is characterized in that the side
Method is after step 1 further include: will according to its similarity of Distance Judgment according to the distance between the vector computing resource of information resources
Similar information resources entity aggregates form a cluster, and different information resources entity division is into different clusters.
8. the information resources of knowledge based map according to claim 1 inquire recommended method, which is characterized in that the step
The similarity between two resources is calculated by COS distance in rapid 4, information resources are calculated by COS distance in the step 5
Similarity between user interest.
9. a kind of information resources of knowledge based map inquire recommender system, which is characterized in that the system comprises:
Data preprocessing module is led to for indicating that knowledge mapping is embedded in low-dimensional vector space by learning model using knowledge mapping
The vectorization that overfitting obtains entity, relationship and attribute indicates；
User interest model constructs module, for analyzing user behavior, understands the interest of user, constructs user interest mould
Type；And
Recommending module is inquired, for the resource acquisition candidate resource collection according to user input query, concentrates and screens in candidate resource
The resource for interest of being close to the users out is recommended.
10. the information resources of knowledge based map according to claim 9 inquire recommender system, which is characterized in that described
Data preprocessing module is also used to the distance between the vector computing resource according to information resources, according to its similarity of Distance Judgment,
Similar information resources entity aggregates are formed into a cluster, different information resources entity division is into different clusters.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910481291.3A CN110188208B (en) | 2019-06-04 | 2019-06-04 | Knowledge graph-based information resource query recommendation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910481291.3A CN110188208B (en) | 2019-06-04 | 2019-06-04 | Knowledge graph-based information resource query recommendation method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110188208A true CN110188208A (en) | 2019-08-30 |
| CN110188208B CN110188208B (en) | 2021-01-26 |
Family
ID=67720192
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910481291.3A Active CN110188208B (en) | 2019-06-04 | 2019-06-04 | Knowledge graph-based information resource query recommendation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110188208B (en) |
Cited By (36)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107704480A (en) * | 2016-08-08 | 2018-02-16 | 百度（美国）有限责任公司 | Extension and the method and system and computer media for strengthening knowledge graph |
| CN110598105A (en) * | 2019-09-10 | 2019-12-20 | 腾讯科技（深圳）有限公司 | Recommendation method, device, equipment and medium based on probability sampling |
| CN110825977A (en) * | 2019-10-10 | 2020-02-21 | 平安科技（深圳）有限公司 | Data recommendation method and related equipment |
| CN110866181A (en) * | 2019-10-12 | 2020-03-06 | 平安国际智慧城市科技股份有限公司 | Resource recommendation method, device and storage medium |
| CN110909124A (en) * | 2019-10-24 | 2020-03-24 | 北京邮电大学 | Hybrid enhanced intelligent demand accurate sensing method and system based on human-in-loop |
| CN110990710A (en) * | 2019-12-24 | 2020-04-10 | 北京百度网讯科技有限公司 | Resource recommendation method and device |
| CN111143694A (en) * | 2019-12-31 | 2020-05-12 | 沈阳航空航天大学 | Information pushing method and device, storage equipment and program |
| CN111191821A (en) * | 2019-12-17 | 2020-05-22 | 东华大学 | Equipment resource allocation optimization method based on knowledge graph drive |
| CN111310051A (en) * | 2019-12-13 | 2020-06-19 | 北京邮电大学 | Cold start solution based on knowledge graph |
| CN111310032A (en) * | 2020-01-21 | 2020-06-19 | 北京健康之家科技有限公司 | Resource recommendation method and device, computer equipment and readable storage medium |
| CN111507796A (en) * | 2020-04-07 | 2020-08-07 | 桂林电子科技大学 | Online shopping mall commodity recommendation method based on knowledge graph |
| CN111538827A (en) * | 2020-04-28 | 2020-08-14 | 清华大学 | Case recommendation method and device based on content and graph neural network and storage medium |
| CN111625707A (en) * | 2020-05-29 | 2020-09-04 | 北京字节跳动网络技术有限公司 | Recommendation response method, device, medium and equipment |
| CN111651678A (en) * | 2020-06-18 | 2020-09-11 | 达而观信息科技（上海）有限公司 | Knowledge graph-based personalized recommendation method |
| CN111813918A (en) * | 2020-06-18 | 2020-10-23 | 国网上海市电力公司 | Scientific and technological resource recommendation processing method and device |
| CN111859125A (en) * | 2020-07-09 | 2020-10-30 | 威海天鑫现代服务技术研究院有限公司 | Semantic network construction and service recommendation method oriented to intellectual property technical resource field |
| CN111931069A (en) * | 2020-09-25 | 2020-11-13 | 浙江口碑网络技术有限公司 | User interest determination method and device and computer equipment |
| CN112052389A (en) * | 2020-08-27 | 2020-12-08 | 安徽聚戎科技信息咨询有限公司 | Knowledge recommendation method based on regional chain |
| CN112115379A (en) * | 2020-09-18 | 2020-12-22 | 吉林农业大学 | Rice variety selection method and device based on knowledge graph |
| CN112131490A (en) * | 2020-09-18 | 2020-12-25 | 东南大学 | Region-sensitive interest point recommendation method driven by knowledge graph |
| CN112559764A (en) * | 2020-12-10 | 2021-03-26 | 北京中视广信科技有限公司 | Content recommendation method based on domain knowledge graph |
| CN112732927A (en) * | 2020-12-31 | 2021-04-30 | 平安资产管理有限责任公司 | Content similarity analysis method and device based on knowledge graph |
| CN112765374A (en) * | 2020-07-27 | 2021-05-07 | 上海斐杰教育科技有限公司 | Education resource screening system and method for information push |
| CN112800326A (en) * | 2021-01-18 | 2021-05-14 | 吉林大学 | Improved Ripp-MKR recommendation method combining multitask learning and knowledge graph |
| CN112818222A (en) * | 2021-01-26 | 2021-05-18 | 吾征智能技术(北京)有限公司 | Knowledge graph-based personalized diet recommendation method and system |
| CN112948417A (en) * | 2021-01-29 | 2021-06-11 | 上海维外科技有限公司 | Sample knowledge polymerization degree calculation method and system based on selection interaction and storage medium |
| CN112966091A (en) * | 2019-12-12 | 2021-06-15 | 复旦大学 | Knowledge graph recommendation system fusing entity information and heat |
| CN113065062A (en) * | 2021-03-09 | 2021-07-02 | 北京工业大学 | News recommendation method and system based on user reading time behavior |
| CN113076478A (en) * | 2021-04-14 | 2021-07-06 | 同济大学 | Technical resource and service recommendation system based on hybrid recommendation algorithm |
| CN113190593A (en) * | 2021-05-12 | 2021-07-30 | 《中国学术期刊（光盘版）》电子杂志社有限公司 | Search recommendation method based on digital human knowledge graph |
| CN113254671A (en) * | 2021-06-22 | 2021-08-13 | 平安科技（深圳）有限公司 | Atlas optimization method, device, equipment and medium based on query analysis |
| CN113742474A (en) * | 2021-11-08 | 2021-12-03 | 北京博瑞彤芸科技股份有限公司 | Intelligent question and answer method and device based on knowledge graph |
| CN113806558A (en) * | 2021-09-22 | 2021-12-17 | 湖北天天数链技术有限公司 | Question selection method, knowledge graph construction method and device and electronic equipment |
| CN116089624A (en) * | 2022-11-17 | 2023-05-09 | 昆仑数智科技有限责任公司 | Knowledge graph-based data recommendation method, device and system |
| CN116235165A (en) * | 2020-08-14 | 2023-06-06 | 西门子股份公司 | Method and device for intelligently providing recommendation information |
| CN117131288A (en) * | 2023-09-07 | 2023-11-28 | 苏州椿之硕信息科技有限公司 | Recommendation system and method for obstetric and academic research |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070239405A1 (en) * | 2004-09-01 | 2007-10-11 | Behrens Clifford A | System and method for consensus-based knowledge validation, analysis and collaboration |
| CN107346346A (en) * | 2017-08-26 | 2017-11-14 | 海南大学 | Learner competencies modeling and learning process Optimal Management System based on data collection of illustrative plates, Information Atlas and knowledge mapping |
| CN107885760A (en) * | 2016-12-21 | 2018-04-06 | 桂林电子科技大学 | It is a kind of to represent learning method based on a variety of semantic knowledge mappings |
| CN108052672A (en) * | 2017-12-29 | 2018-05-18 | 北京师范大学 | Promote structural knowledge map construction system and method using group study behavior |
| CN108733798A (en) * | 2018-05-17 | 2018-11-02 | 电子科技大学 | A kind of personalized recommendation method of knowledge based collection of illustrative plates |
| CN109213872A (en) * | 2018-09-11 | 2019-01-15 | 中国电子科技集团公司第二十八研究所 | Knowledge based indicates the entity relationship prediction technique and forecasting system of study |
-
2019
- 2019-06-04 CN CN201910481291.3A patent/CN110188208B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070239405A1 (en) * | 2004-09-01 | 2007-10-11 | Behrens Clifford A | System and method for consensus-based knowledge validation, analysis and collaboration |
| CN107885760A (en) * | 2016-12-21 | 2018-04-06 | 桂林电子科技大学 | It is a kind of to represent learning method based on a variety of semantic knowledge mappings |
| CN107346346A (en) * | 2017-08-26 | 2017-11-14 | 海南大学 | Learner competencies modeling and learning process Optimal Management System based on data collection of illustrative plates, Information Atlas and knowledge mapping |
| CN108052672A (en) * | 2017-12-29 | 2018-05-18 | 北京师范大学 | Promote structural knowledge map construction system and method using group study behavior |
| CN108733798A (en) * | 2018-05-17 | 2018-11-02 | 电子科技大学 | A kind of personalized recommendation method of knowledge based collection of illustrative plates |
| CN109213872A (en) * | 2018-09-11 | 2019-01-15 | 中国电子科技集团公司第二十八研究所 | Knowledge based indicates the entity relationship prediction technique and forecasting system of study |
Cited By (50)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107704480A (en) * | 2016-08-08 | 2018-02-16 | 百度（美国）有限责任公司 | Extension and the method and system and computer media for strengthening knowledge graph |
| CN110598105A (en) * | 2019-09-10 | 2019-12-20 | 腾讯科技（深圳）有限公司 | Recommendation method, device, equipment and medium based on probability sampling |
| CN110598105B (en) * | 2019-09-10 | 2021-10-08 | 腾讯科技（深圳）有限公司 | Recommendation method, device, equipment and medium based on probability sampling |
| CN110825977A (en) * | 2019-10-10 | 2020-02-21 | 平安科技（深圳）有限公司 | Data recommendation method and related equipment |
| CN110866181A (en) * | 2019-10-12 | 2020-03-06 | 平安国际智慧城市科技股份有限公司 | Resource recommendation method, device and storage medium |
| CN110866181B (en) * | 2019-10-12 | 2022-04-22 | 平安国际智慧城市科技股份有限公司 | Resource recommendation method, device and storage medium |
| CN110909124A (en) * | 2019-10-24 | 2020-03-24 | 北京邮电大学 | Hybrid enhanced intelligent demand accurate sensing method and system based on human-in-loop |
| CN112966091B (en) * | 2019-12-12 | 2022-10-14 | 复旦大学 | Knowledge map recommendation system fusing entity information and heat |
| CN112966091A (en) * | 2019-12-12 | 2021-06-15 | 复旦大学 | Knowledge graph recommendation system fusing entity information and heat |
| CN111310051A (en) * | 2019-12-13 | 2020-06-19 | 北京邮电大学 | Cold start solution based on knowledge graph |
| CN111191821A (en) * | 2019-12-17 | 2020-05-22 | 东华大学 | Equipment resource allocation optimization method based on knowledge graph drive |
| CN110990710A (en) * | 2019-12-24 | 2020-04-10 | 北京百度网讯科技有限公司 | Resource recommendation method and device |
| CN111143694A (en) * | 2019-12-31 | 2020-05-12 | 沈阳航空航天大学 | Information pushing method and device, storage equipment and program |
| CN111143694B (en) * | 2019-12-31 | 2023-06-16 | 沈阳航空航天大学 | Information pushing method and device, storage device and program |
| CN111310032A (en) * | 2020-01-21 | 2020-06-19 | 北京健康之家科技有限公司 | Resource recommendation method and device, computer equipment and readable storage medium |
| CN111310032B (en) * | 2020-01-21 | 2023-09-22 | 北京水滴科技集团有限公司 | Resource recommendation method, device, computer equipment and readable storage medium |
| CN111507796A (en) * | 2020-04-07 | 2020-08-07 | 桂林电子科技大学 | Online shopping mall commodity recommendation method based on knowledge graph |
| CN111538827A (en) * | 2020-04-28 | 2020-08-14 | 清华大学 | Case recommendation method and device based on content and graph neural network and storage medium |
| CN111538827B (en) * | 2020-04-28 | 2023-09-05 | 清华大学 | Case recommendation method, device and storage medium based on content and graph neural network |
| CN111625707A (en) * | 2020-05-29 | 2020-09-04 | 北京字节跳动网络技术有限公司 | Recommendation response method, device, medium and equipment |
| CN111625707B (en) * | 2020-05-29 | 2023-04-14 | 北京字节跳动网络技术有限公司 | Recommendation response method, device, medium and equipment |
| CN111651678B (en) * | 2020-06-18 | 2023-12-22 | 达观数据有限公司 | Personalized recommendation method based on knowledge graph |
| CN111813918A (en) * | 2020-06-18 | 2020-10-23 | 国网上海市电力公司 | Scientific and technological resource recommendation processing method and device |
| CN111651678A (en) * | 2020-06-18 | 2020-09-11 | 达而观信息科技（上海）有限公司 | Knowledge graph-based personalized recommendation method |
| CN111859125A (en) * | 2020-07-09 | 2020-10-30 | 威海天鑫现代服务技术研究院有限公司 | Semantic network construction and service recommendation method oriented to intellectual property technical resource field |
| CN112765374A (en) * | 2020-07-27 | 2021-05-07 | 上海斐杰教育科技有限公司 | Education resource screening system and method for information push |
| CN116235165A (en) * | 2020-08-14 | 2023-06-06 | 西门子股份公司 | Method and device for intelligently providing recommendation information |
| CN112052389A (en) * | 2020-08-27 | 2020-12-08 | 安徽聚戎科技信息咨询有限公司 | Knowledge recommendation method based on regional chain |
| CN112115379A (en) * | 2020-09-18 | 2020-12-22 | 吉林农业大学 | Rice variety selection method and device based on knowledge graph |
| CN112131490A (en) * | 2020-09-18 | 2020-12-25 | 东南大学 | Region-sensitive interest point recommendation method driven by knowledge graph |
| CN111931069A (en) * | 2020-09-25 | 2020-11-13 | 浙江口碑网络技术有限公司 | User interest determination method and device and computer equipment |
| CN112559764B (en) * | 2020-12-10 | 2023-12-01 | 北京中视广信科技有限公司 | Content recommendation method based on domain knowledge graph |
| CN112559764A (en) * | 2020-12-10 | 2021-03-26 | 北京中视广信科技有限公司 | Content recommendation method based on domain knowledge graph |
| CN112732927A (en) * | 2020-12-31 | 2021-04-30 | 平安资产管理有限责任公司 | Content similarity analysis method and device based on knowledge graph |
| CN112800326A (en) * | 2021-01-18 | 2021-05-14 | 吉林大学 | Improved Ripp-MKR recommendation method combining multitask learning and knowledge graph |
| CN112800326B (en) * | 2021-01-18 | 2022-03-15 | 吉林大学 | Improved Ripp-MKR recommendation method combining multitask learning and knowledge graph |
| CN112818222A (en) * | 2021-01-26 | 2021-05-18 | 吾征智能技术(北京)有限公司 | Knowledge graph-based personalized diet recommendation method and system |
| CN112818222B (en) * | 2021-01-26 | 2024-02-23 | 吾征智能技术(北京)有限公司 | Personalized diet recommendation method and system based on knowledge graph |
| CN112948417A (en) * | 2021-01-29 | 2021-06-11 | 上海维外科技有限公司 | Sample knowledge polymerization degree calculation method and system based on selection interaction and storage medium |
| CN113065062A (en) * | 2021-03-09 | 2021-07-02 | 北京工业大学 | News recommendation method and system based on user reading time behavior |
| CN113076478A (en) * | 2021-04-14 | 2021-07-06 | 同济大学 | Technical resource and service recommendation system based on hybrid recommendation algorithm |
| CN113190593A (en) * | 2021-05-12 | 2021-07-30 | 《中国学术期刊（光盘版）》电子杂志社有限公司 | Search recommendation method based on digital human knowledge graph |
| CN113254671B (en) * | 2021-06-22 | 2021-09-28 | 平安科技（深圳）有限公司 | Atlas optimization method, device, equipment and medium based on query analysis |
| CN113254671A (en) * | 2021-06-22 | 2021-08-13 | 平安科技（深圳）有限公司 | Atlas optimization method, device, equipment and medium based on query analysis |
| CN113806558A (en) * | 2021-09-22 | 2021-12-17 | 湖北天天数链技术有限公司 | Question selection method, knowledge graph construction method and device and electronic equipment |
| CN113806558B (en) * | 2021-09-22 | 2024-03-26 | 湖北天天数链技术有限公司 | Question selection method, knowledge graph construction device and electronic equipment |
| CN113742474A (en) * | 2021-11-08 | 2021-12-03 | 北京博瑞彤芸科技股份有限公司 | Intelligent question and answer method and device based on knowledge graph |
| CN116089624A (en) * | 2022-11-17 | 2023-05-09 | 昆仑数智科技有限责任公司 | Knowledge graph-based data recommendation method, device and system |
| CN116089624B (en) * | 2022-11-17 | 2024-02-27 | 昆仑数智科技有限责任公司 | Knowledge graph-based data recommendation method, device and system |
| CN117131288A (en) * | 2023-09-07 | 2023-11-28 | 苏州椿之硕信息科技有限公司 | Recommendation system and method for obstetric and academic research |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110188208B (en) | 2021-01-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110188208A (en) | A kind of the information resources inquiry recommended method and system of knowledge based map | |
| CN112214685B (en) | Knowledge graph-based personalized recommendation method | |
| Yin et al. | LCARS: A spatial item recommender system | |
| Berkovsky et al. | Mediation of user models for enhanced personalization in recommender systems | |
| CN110413780A (en) | Text emotion analysis method, device, storage medium and electronic equipment | |
| CN108446964B (en) | User recommendation method based on mobile traffic DPI data | |
| CN101111837A (en) | Search processing with automatic categorization of queries | |
| CN101834837A (en) | On-line landscape video active information service system of scenic spots in tourist attraction based on bandwidth network | |
| CN102663001A (en) | Automatic blog writer interest and character identifying method based on support vector machine | |
| Schickel-Zuber et al. | Using hierarchical clustering for learning theontologies used in recommendation systems | |
| Xie et al. | Generative adversarial network based service recommendation in heterogeneous information networks | |
| CN104572733B (en) | The method and device of user interest labeling | |
| Liu et al. | Using collaborative filtering algorithms combined with Doc2Vec for movie recommendation | |
| CN107545075B (en) | Restaurant recommendation method based on online comments and context awareness | |
| CN103095849B (en) | A method and a system of spervised web service finding based on attribution forecast and error correction of quality of service (QoS) | |
| Zhang et al. | Application and analysis of artificial intelligence in college students’ career planning and employment and entrepreneurship information recommendation | |
| CN116823410B (en) | Data processing method, object processing method, recommending method and computing device | |
| Zheng | Multi level recommendation system of college online learning resources based on multi intelligence algorithm | |
| Jaffali et al. | Survey on social networks data analysis | |
| CN112784160B (en) | Interest point recommendation method based on tensor decomposition and considering context semantic information | |
| CN115018255A (en) | Tourist attraction evaluation information quality validity analysis method based on integrated learning data mining technology | |
| CN112069390B (en) | User book borrowing behavior analysis and interest prediction method based on space-time dimension | |
| Quadrana | Algorithms for sequence-aware recommender systems | |
| Ma et al. | Deep learning framework for online interactive service recommendation in iterative mashup development | |
| Duan et al. | PBEM: A pattern-based embedding model for user location category prediction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |