CN113742594B - Recommendation system recall method and device - Google Patents
Recommendation system recall method and device Download PDFInfo
- Publication number
- CN113742594B CN113742594B CN202111088806.7A CN202111088806A CN113742594B CN 113742594 B CN113742594 B CN 113742594B CN 202111088806 A CN202111088806 A CN 202111088806A CN 113742594 B CN113742594 B CN 113742594B
- Authority
- CN
- China
- Prior art keywords
- user
- vector
- features
- project
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 35
- 239000013598 vector Substances 0.000 claims abstract description 526
- 238000013135 deep learning Methods 0.000 claims abstract description 82
- 238000004364 calculation method Methods 0.000 claims abstract description 69
- 238000013528 artificial neural network Methods 0.000 claims abstract description 29
- 238000009825 accumulation Methods 0.000 claims description 37
- 238000012545 processing Methods 0.000 claims description 26
- 238000004590 computer program Methods 0.000 claims description 16
- 230000006870 function Effects 0.000 claims description 13
- 238000005094 computer simulation Methods 0.000 claims description 5
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 238000013136 deep learning model Methods 0.000 description 25
- 238000010586 diagram Methods 0.000 description 11
- 238000011161 development Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000012163 sequencing technique Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 101100481876 Danio rerio pbk gene Proteins 0.000 description 1
- 101100481878 Mus musculus Pbk gene Proteins 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000001568 sexual effect Effects 0.000 description 1
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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a recall method and device of a recommendation system, and relates to the technical field of artificial intelligence, wherein the method comprises the following steps: discretizing and vectorizing the user features and the project features respectively, inputting the user features and the project features into a multi-layer fully-connected neural network respectively to obtain deep learning output vectors, and accumulating the discrete project features of the user features and the project features to obtain linear model output vectors; splicing the deep learning output vector and the linear model output vector to obtain an output total vector, carrying out dot product multiplication on the output total vector of the user features and the output total vector of the item features to obtain a calculation model of the matching degree between the user features and the item features, and calculating the matching degree between the user features of the recall target user and the item features of the recall target item; and carrying out recall operation according to the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item. The invention can improve the recall accuracy and recall efficiency of the recommendation system.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a method and a device for recommending system recall.
Background
This section is intended to provide a background or context to the embodiments of the invention that are recited in the claims. The description herein is not admitted to be prior art by inclusion in this section.
With the development of society and the popularization of the Internet, the information amount is increased faster and faster, information searching is changed from people to information searching, and a recommendation system is intelligent and personalized, extracts useful information from mass information and actively pushes the useful information to people.
The recommendation system needs to find matching information from massive data information to carry out personalized pushing, and in consideration of time cost, the recommendation system generally comprises two links of recall and sequencing, and the recall link is responsible for quickly finding n pieces of information which are possibly matched from the massive information. Vector retrieval is one way to recall links.
Since the current deep learning model for recommender recall is typically a dual-tower architecture, the dual-tower architecture outputs are in the form of vectors, which are then retrieved. The vector form limits the capability of the deep learning model, and some more complex models cannot be used, so that the model capability of the recall of the recommendation system is limited, and the recall accuracy of the recommendation system is reduced; meanwhile, the current deep learning model for recall of the recommendation system cannot calculate linear feature vectors, so that recall efficiency of the recommendation system is reduced.
Disclosure of Invention
The embodiment of the invention provides a recall method of a recommendation system, which is used for improving the recall accuracy and the work efficiency of the recommendation system, and comprises the following steps:
discretizing and vectorizing the user features to obtain a plurality of user discrete feature vectors of the user features; inputting a plurality of user discrete feature vectors of the user features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the user features; vector accumulation is carried out on a plurality of user discrete feature vectors of the user features to obtain linear output vectors of the user features; vector splicing is carried out on the deep learning output vector and the linear output vector of the user characteristic, and an output total vector of the user characteristic is obtained;
discretizing and vectorizing the project features to obtain a plurality of project discrete feature vectors of the project features; inputting a plurality of project discrete feature vectors of project features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the project features; vector accumulation is carried out on a plurality of project discrete feature vectors of the project features, so that linear output vectors of the project features are obtained; vector splicing is carried out on the deep learning output vector and the linear output vector of the project feature, and an output total vector of the project feature is obtained;
Multiplying the output total vector of the user features and the output total vector of the project features by dot products to obtain a calculation model of the matching degree between the user features and the project features;
inputting the user characteristics of the recall target user and the item characteristics of the recall target item into the calculation model to obtain the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item;
and carrying out recall operation according to the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item.
The embodiment of the invention also provides a device for recalling the recommendation system, which is used for improving the accuracy and the working efficiency of the recall of the recommendation system, and comprises the following components:
the user feature processing module is used for carrying out discretization and vectorization processing on the user features to obtain a plurality of user discrete feature vectors of the user features; inputting a plurality of user discrete feature vectors of the user features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the user features; vector accumulation is carried out on a plurality of user discrete feature vectors of the user features to obtain linear output vectors of the user features; vector splicing is carried out on the deep learning output vector and the linear output vector of the user characteristic, and an output total vector of the user characteristic is obtained;
The project feature processing module is used for carrying out discretization and vectorization processing on project features to obtain a plurality of project discrete feature vectors of the project features; inputting a plurality of project discrete feature vectors of project features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the project features; vector accumulation is carried out on a plurality of project discrete feature vectors of the project features, so that linear output vectors of the project features are obtained; vector splicing is carried out on the deep learning output vector and the linear output vector of the project feature, and an output total vector of the project feature is obtained;
the computing model modeling module is used for multiplying the output total vector of the user features and the output total vector of the project features by dot products to obtain a computing model of the matching degree between the user features and the project features;
the matching degree calculation module is used for inputting the user characteristics of the recall target user and the item characteristics of the recall target item into the calculation model to obtain the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item;
and the recommendation system recall module is used for carrying out recall operation according to the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item.
The embodiment of the invention also provides computer equipment, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the method for recalling the recommendation system is realized when the processor executes the computer program.
The embodiment of the invention also provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program for executing the recall method of the recommendation system.
In the embodiment of the invention, discretization and vectorization are carried out on the user characteristics to obtain a plurality of user discrete characteristic vectors of the user characteristics; inputting a plurality of user discrete feature vectors of the user features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the user features; vector accumulation is carried out on a plurality of user discrete feature vectors of the user features to obtain linear output vectors of the user features; vector splicing is carried out on the deep learning output vector and the linear output vector of the user characteristic, and an output total vector of the user characteristic is obtained; discretizing and vectorizing the project features to obtain a plurality of project discrete feature vectors of the project features; inputting a plurality of project discrete feature vectors of project features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the project features; vector accumulation is carried out on a plurality of project discrete feature vectors of the project features, so that linear output vectors of the project features are obtained; vector splicing is carried out on the deep learning output vector and the linear output vector of the project feature, and an output total vector of the project feature is obtained; multiplying the output total vector of the user features and the output total vector of the project features by dot products to obtain a calculation model of the matching degree between the user features and the project features; inputting the user characteristics of the recall target user and the item characteristics of the recall target item into the calculation model to obtain the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item; according to the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item, recall operation is carried out, so that the accurate calculation of the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item can be realized by establishing a calculation model of the matching degree between the user characteristics and the item characteristics, the calculation of linear discrete characteristic vectors can be realized by determining linear output vectors, the problem that the recall capability of a recommendation system is limited because the traditional deep learning model can only carry out vector form calculation in the prior art is avoided, the aim of carrying out explicit second-order intersection of each characteristic on the discrete characteristic vectors can be realized by linearly outputting the vectors, and the recall accuracy of the recommendation system is improved; meanwhile, through linear output vectors, the recall of the recommendation system can be enabled to have memory capacity based on explicit feature cross combination, and recall efficiency of the recommendation system is improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art. In the drawings:
FIG. 1 is a flowchart of a method for recommending system recall according to an embodiment of the present invention;
FIG. 2 is a diagram illustrating a method for recommending system recall according to an embodiment of the present invention;
FIG. 3 is a schematic diagram illustrating a recall device of a recommendation system according to an embodiment of the present invention;
FIG. 4 is a diagram illustrating an exemplary apparatus for recommending system recall according to an embodiment of the present invention;
FIG. 5 is a schematic diagram of a computer device for recommending system recall in an embodiment of the invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the embodiments of the present invention will be described in further detail with reference to the accompanying drawings. The exemplary embodiments of the present invention and their descriptions herein are for the purpose of explaining the present invention, but are not to be construed as limiting the invention.
Along with the development of society and popularization of the Internet, the information amount is increased more and more rapidly, information searching is changed from people to information searching, and a recommendation system is intelligent and personalized, extracts useful information from massive information and actively pushes the useful information to people.
The recommendation system needs to find matching information from massive data information to perform personalized pushing, and in consideration of time cost, the recommendation system generally comprises two links of recall and sequencing, and the recall link is responsible for quickly finding n pieces of information which are possibly matched from the massive information. Vector retrieval is one way to recall links.
Currently, vector retrieval requires that User and Item be separately characterized as vectors, respectively, but this form limits the use of more complex deep learning models in the recall step. In the ranking link of the recommendation system, the deep FM model is widely used, and the deep FM model is combined with the deep learning model and the FM model, so that the deep FM model has the generalization capability of deep learning and the memory capability of a linear model, and the model performance is enhanced. However, since the recall link of the recommendation system is limited to be a vector, the model cannot be directly used for the recall link of the recommendation system. The current deep learning model for recommender recall is typically a double-tower architecture, where the output is in the form of a vector, which is then retrieved. The vector form limits the ability of the deep learning model, and some more complex models (such as deep fm models, which are commonly used in recommender ranking links, cannot be used for recommender recall) cannot be used, so the recommender recall model is limited in ability. The existing scheme only uses a deep learning model and has stronger generalization capability, but the architecture can not be combined with a linear model to strengthen the memory capability.
In summary, since the current deep learning model for recommender recall is typically a dual-tower architecture, the dual-tower architecture outputs a vector in the form of a vector, which is then retrieved. The vector form limits the capability of the deep learning model, and some more complex models cannot be used, so that the model capability of the recall of the recommendation system is limited, and the recall accuracy of the recommendation system is reduced; meanwhile, the current deep learning model for recall of the recommendation system cannot calculate linear feature vectors, so that recall efficiency of the recommendation system is reduced.
In order to solve the above problems, an embodiment of the present invention provides a method for recalling a recommendation system, which is used to improve accuracy and working efficiency of the recall of the recommendation system, as shown in fig. 1, the method may include:
step 101: discretizing and vectorizing the user features to obtain a plurality of user discrete feature vectors of the user features; inputting a plurality of user discrete feature vectors of the user features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the user features; vector accumulation is carried out on a plurality of user discrete feature vectors of the user features to obtain linear output vectors of the user features; vector splicing is carried out on the deep learning output vector and the linear output vector of the user characteristic, and an output total vector of the user characteristic is obtained;
Step 102: discretizing and vectorizing the project features to obtain a plurality of project discrete feature vectors of the project features; inputting a plurality of project discrete feature vectors of project features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the project features; vector accumulation is carried out on a plurality of project discrete feature vectors of the project features, so that linear output vectors of the project features are obtained; vector splicing is carried out on the deep learning output vector and the linear output vector of the project feature, and an output total vector of the project feature is obtained;
step 103: multiplying the output total vector of the user features and the output total vector of the project features by dot products to obtain a calculation model of the matching degree between the user features and the project features;
step 104: inputting the user characteristics of the recall target user and the item characteristics of the recall target item into the calculation model to obtain the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item;
step 105: and carrying out recall operation according to the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item.
In the embodiment of the invention, discretization and vectorization are carried out on the user characteristics to obtain a plurality of user discrete characteristic vectors of the user characteristics; inputting a plurality of user discrete feature vectors of the user features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the user features; vector accumulation is carried out on a plurality of user discrete feature vectors of the user features to obtain linear output vectors of the user features; vector splicing is carried out on the deep learning output vector and the linear output vector of the user characteristic, and an output total vector of the user characteristic is obtained; discretizing and vectorizing the project features to obtain a plurality of project discrete feature vectors of the project features; inputting a plurality of project discrete feature vectors of project features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the project features; vector accumulation is carried out on a plurality of project discrete feature vectors of the project features, so that linear output vectors of the project features are obtained; vector splicing is carried out on the deep learning output vector and the linear output vector of the project feature, and an output total vector of the project feature is obtained; multiplying the output total vector of the user features and the output total vector of the project features by dot products to obtain a calculation model of the matching degree between the user features and the project features; inputting the user characteristics of the recall target user and the item characteristics of the recall target item into the calculation model to obtain the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item; according to the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item, recall operation is carried out, so that the accurate calculation of the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item can be realized by establishing a calculation model of the matching degree between the user characteristics and the item characteristics, the calculation of linear discrete characteristic vectors can be realized by determining linear output vectors, the problem that the recall capability of a recommendation system is limited because the traditional deep learning model can only carry out vector form calculation in the prior art is avoided, the aim of carrying out explicit second-order intersection of each characteristic on the discrete characteristic vectors can be realized by linearly outputting the vectors, and the recall accuracy of the recommendation system is improved; meanwhile, through linear output vectors, the recall of the recommendation system can be enabled to have memory capacity based on explicit feature cross combination, and recall efficiency of the recommendation system is improved.
When the method is implemented, discretization and vectorization processing are carried out on the user characteristics to obtain a plurality of user discrete characteristic vectors of the user characteristics; inputting a plurality of user discrete feature vectors of the user features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the user features; vector accumulation is carried out on a plurality of user discrete feature vectors of the user features to obtain linear output vectors of the user features; and vector splicing is carried out on the deep learning output vector and the linear output vector of the user characteristic, so that the total output vector of the user characteristic is obtained.
In one embodiment, vector accumulation may be performed on a plurality of user discrete feature vectors of the user feature to obtain a linear output vector of the user feature according to the following formula:
wherein U is FM A linear output vector representing a user characteristic; u (U) i Representing an ith user discrete feature vector; m represents the number of discrete feature vectors of the user.
In one embodiment, the deep learning output vector and the linear output vector of the user feature may be vector spliced according to the following formula to obtain the total output vector of the user feature:
wherein U is FM A linear output vector representing a user characteristic; u (U) Deep A deep learning output vector representing a user feature; u represents the output total vector of the user features; Representing vector concatenation operations.
In the embodiment, the linear user discrete feature vector can be calculated by determining the linear output vector of the user feature, so that the problem that the recall capability of the recommendation system is limited because the traditional deep learning model can only perform vector form calculation in the prior art is avoided, the aim of performing explicit second-order intersection of each feature on the user discrete feature vector can be fulfilled through the linear output vector of the user feature, and the recall accuracy of the recommendation system is improved; meanwhile, through the linear output vector of the user characteristics, the recall of the recommendation system can be enabled to have the memory capacity based on the explicit characteristic cross combination, and the recall efficiency of the recommendation system is improved.
When the method is implemented, discretization and vectorization processing are carried out on the user characteristics, so that a plurality of user discrete characteristic vectors of the user characteristics are obtained; inputting a plurality of user discrete feature vectors of the user features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the user features; vector accumulation is carried out on a plurality of user discrete feature vectors of the user features to obtain linear output vectors of the user features; vector splicing is carried out on the deep learning output vector and the linear output vector of the user feature, after the output total vector of the user feature is obtained, discretization and vectorization are carried out on the project feature, and a plurality of project discrete feature vectors of the project feature are obtained; inputting a plurality of project discrete feature vectors of project features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the project features; vector accumulation is carried out on a plurality of project discrete feature vectors of the project features, so that linear output vectors of the project features are obtained; and vector splicing is carried out on the deep learning output vector and the linear output vector of the project feature, so that the total output vector of the project feature is obtained.
In one embodiment, the plurality of item discrete feature vectors of the item feature may be vector accumulated to obtain a linear output vector of the item feature according to the following formula:
wherein I is FM A linear output vector representing the item feature; i j Representing a j-th item discrete feature vector; n represents the number of discrete feature vectors of the item.
In one embodiment, the deep learning output vector and the linear output vector of the project feature may be vector spliced according to the following formula to obtain the output total vector of the project feature:
wherein I is FM A linear output vector representing the item feature; i Deep A deep learning output vector representing project features; i respectively represents the output total vector of project characteristics;representing vector concatenation operations.
In the embodiment, the linear project discrete feature vector can be calculated by determining the linear output vector of the project feature, so that the problem that recall capability of a recommendation system is limited because a traditional deep learning model can only perform vector form calculation in the prior art is avoided, the aim of performing explicit second-order intersection of each feature on the project discrete feature vector can be fulfilled through the linear output vector of the project feature, and recall accuracy of the recommendation system is improved; meanwhile, through the linear output vector of the project features, the recall of the recommendation system can be enabled to have the memory capacity based on explicit feature cross combination, and the recall efficiency of the recommendation system is improved.
When the method is implemented, discretization and vectorization are carried out on project features to obtain a plurality of project discrete feature vectors of the project features; inputting a plurality of project discrete feature vectors of project features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the project features; vector accumulation is carried out on a plurality of project discrete feature vectors of the project features, so that linear output vectors of the project features are obtained; and vector splicing is carried out on the deep learning output vector and the linear output vector of the project feature, after the output total vector of the project feature is obtained, dot product multiplication is carried out on the output total vector of the user feature and the output total vector of the project feature, and a calculation model of the matching degree between the user feature and the project feature is obtained.
In one embodiment, the dot product multiplication can be performed on the output total vector of the user feature and the output total vector of the item feature according to the following formula, so as to obtain a calculation model of the matching degree between the user feature and the item feature:
wherein Score represents a computational model of the degree of matching between user features and project features; u (U) i And I i Respectively representing an ith user discrete feature vector and a jth item discrete feature vector; m and n represent the number of user discrete feature vectors and project discrete feature vectors, respectively; Wherein U is FM And I FM A linear output vector representing the user feature and a linear output vector representing the project feature, respectively, U Deep And I Deep The deep learning output vector and the deep learning output vector of the project feature respectively represent the user feature, and U and I represent the total output vector of the user feature and the total output vector of the project feature respectively, +.>Representing a vector concatenation operation; /> Wherein U is i And I j Respectively representing an ith user discrete feature vector and a jth item discrete feature vector.
In the above embodiment, by establishing the calculation model of the matching degree between the user feature and the item feature, accurate calculation of the matching degree between the user feature of the recall target user and the item feature of the recall target item is achieved, and FM (factorization machine, FM model is a matrix-decomposition-based machine learning model, that is, a linear model) is obtained by the accumulation and summation operation of each discrete feature, and the operation of splicing the accumulation and summation vector, the deep learning output vector and the FM accumulation and summation vector is achieved because the two steps achieve adding the linear model FM model, and the combination of the deep learning model and the FM model is achieved. According to the calculation model for the matching degree between the user features and the project features, the linear M model is additionally added, and explicit second-order intersection of the features can be performed, so that the model has generalization capability of a deep learning model and memory capability based on explicit feature combination and intersection, and performance of the model is improved.
In the specific implementation, after dot product multiplication is carried out on the output total vector of the user features and the output total vector of the item features to obtain a calculation model of the matching degree between the user features and the item features, the user features of the recall target user and the item features of the recall target item are input into the calculation model to obtain the matching degree between the user features of the recall target user and the item features of the recall target item.
In the above embodiment, by combining the user characteristics of the recall target user and the item characteristics of the recall target item, the calculation of the degree of matching between the user characteristics of the recall target user and the item characteristics of the recall target item can be achieved, facilitating the recall operation in the subsequent step.
In the specific implementation, after the user characteristics of the recall target user and the item characteristics of the recall target item are input into the calculation model to obtain the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item, recall operation is performed according to the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item.
In the embodiment, the combination model of the deep learning model and the linear model FM model is used for the recall link of the recommendation system, wherein the FM model can perform explicit second-order intersection of each feature, so that the model not only has the generalization capability of the deep learning model, but also has the memory capability based on explicit feature intersection combination, thereby improving the performance of the model, and finally improving the recall accuracy and recall efficiency of the recommendation system.
In specific implementation, the method for recalling the recommendation system provided by the embodiment of the invention can further comprise the following steps:
correcting the matching degree calculation model of the user characteristics and the project characteristics according to the cross entropy loss function to obtain a corrected matching degree calculation model of the user characteristics and the project characteristics;
inputting user characteristics of a recall target user and item characteristics of a recall target item into the computing model, including:
and inputting the user characteristics of the recall target user and the item characteristics of the recall target item into a corrected calculation model of the matching degree between the user characteristics and the item characteristics.
In an embodiment, the cross entropy loss function may be a softmax cross entropy loss function.
In the embodiment, the matching degree calculation model of the user features and the project features is corrected through the cross entropy loss function, so that training of the matching degree calculation model between the user features and the project features can be realized, the accuracy of the matching degree calculation model between the user features and the project features is improved, and the recall accuracy and recall efficiency of the recommendation system can be indirectly improved.
In specific implementation, the method for recalling the recommendation system provided by the embodiment of the invention can further comprise the following steps:
And calculating and transmitting the loss amount of the calculation model of the matching degree between the user characteristics and the project characteristics.
In the above embodiment, the loss amount of the calculation model of the matching degree between the user feature and the project feature is calculated and sent, so that the staff can calculate the model precision according to the loss amount of the model.
A specific embodiment is given below, in connection with fig. 2, to illustrate a specific application of the method of the invention, which embodiment may comprise the following steps:
1. discretizing User side features (i.e., the User features described above) and Item side features (i.e., the Item features described above), and then assigning a vector (i.e., the DenseEmbeddings in FIG. 2) to the value of each discretized feature (i.e., the sparseFeatenes and the Item discrete features described above in FIG. 2) to obtain a plurality of User discrete feature vectors and a plurality of Item discrete feature vectors as described above.
2. The vectors corresponding to the discretization characteristics of the User side and the Item side are respectively spliced to be used as the input of the multi-layer fully-connected neural network, and the output vector of the multi-layer fully-connected neural network is used as the output vector (deep learning) of the deep learning part of the model, namely U Deep And I Deep . The multi-layer fully-connected neural network is a deep learning model;
3. at the same time, vectors corresponding to all features of the User side and the Item side are respectively subjected to vector accumulation, and the accumulated result is used as an output vector of the FM part of the model (namely FMEmbeddings in FIG. 2), namely U FM And I FM ：
Wherein U is i And I i Vectors corresponding to the ith discretization feature of the User side and the Item side respectively; m and n represent the number of discrete feature vectors on the User side and Item side, respectively.
4. The output vectors (deep learning) of the User side and the Item side and the linear output vectors (FMEmbeddings) of the FM part are respectively spliced to obtain the model output total vectors of the User side and the Item side, namely
Wherein,representing vector concatenation operations.
5. And carrying out dot product multiplication on the total output vector of the model of the User and the Item to obtain the matching degree score of the User and the Item. The Item side not only contains the discretized feature vector of the positive sample, but also contains the discretized feature vectors of a plurality of negative samples, and the model training can be performed by using the softmax cross entropy loss function.
Matching degree score of User and Item
Wherein,that is, the FM model, i.e., the explicit second-order intersection of the User side and the Item side is implemented in vector multiplication, U Deep ·I Deep Is the result of the deep learning model. Therefore, the model has generalization capability of the deep learning model and memory capability based on explicit feature combination and intersection, so that the performance of the model is improved.
6. And (3) reasoning by using the trained model to obtain vectors corresponding to each User and each Item, and searching in the whole items by using the vector of a User to obtain the TopK best matched items.
In the above example, the vectors corresponding to all the features on the User side and the Item side may be accumulated, respectively, and the accumulated result is used as the output vector of the FM part of the model; the output vectors of the deep learning part and the output vectors of the FM part on the User side and the Item side can be spliced respectively to obtain model output vectors on the User side and the Item side, so that discretization of User features and features of projects is completed, the discretized features are assigned with vectors, and the accumulation and summation operation of each discrete feature, the splicing operation of the deep learning output vectors and the FM accumulation and summation vectors, the Item vector multiplication calculation loss function of the User vectors and positive and negative samples are facilitated through the establishment of a fully connected neural network, and the model is optimized. The two steps of adding the linear model FM model and combining the deep learning model and the FM model are realized through the accumulation summation operation of each discrete feature and the splicing operation of the deep learning output vector and the FM accumulation summation vector.
Of course, it is to be understood that other variations of the above detailed procedures are also possible, and all related variations should fall within the protection scope of the present invention.
In the embodiment of the invention, discretization and vectorization are carried out on the user characteristics to obtain a plurality of user discrete characteristic vectors of the user characteristics; inputting a plurality of user discrete feature vectors of the user features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the user features; vector accumulation is carried out on a plurality of user discrete feature vectors of the user features to obtain linear output vectors of the user features; vector splicing is carried out on the deep learning output vector and the linear output vector of the user characteristic, and an output total vector of the user characteristic is obtained; discretizing and vectorizing the project features to obtain a plurality of project discrete feature vectors of the project features; inputting a plurality of project discrete feature vectors of project features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the project features; vector accumulation is carried out on a plurality of project discrete feature vectors of the project features, so that linear output vectors of the project features are obtained; vector splicing is carried out on the deep learning output vector and the linear output vector of the project feature, and an output total vector of the project feature is obtained; multiplying the output total vector of the user features and the output total vector of the project features by dot products to obtain a calculation model of the matching degree between the user features and the project features; inputting the user characteristics of the recall target user and the item characteristics of the recall target item into the calculation model to obtain the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item; according to the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item, recall operation is carried out, so that the accurate calculation of the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item can be realized by establishing a calculation model of the matching degree between the user characteristics and the item characteristics, the calculation of linear discrete characteristic vectors can be realized by determining linear output vectors, the problem that the recall capability of a recommendation system is limited because the traditional deep learning model can only carry out vector form calculation in the prior art is avoided, the aim of carrying out explicit second-order intersection of each characteristic on the discrete characteristic vectors can be realized by linearly outputting the vectors, and the recall accuracy of the recommendation system is improved; meanwhile, through linear output vectors, the recall of the recommendation system can be enabled to have memory capacity based on explicit feature cross combination, and recall efficiency of the recommendation system is improved.
The embodiment of the invention also provides a device for recommending system recall, and the embodiment is as described in the following. Because the principle of the device for solving the problem is similar to that of the recall method of the recommendation system, the implementation of the device can refer to the implementation of the recall method of the recommendation system, and the repeated parts are not repeated.
The embodiment of the invention also provides a device for recalling the recommendation system, which is used for improving the accuracy and the working efficiency of the recall of the recommendation system, and as shown in fig. 3, the device can comprise:
the user characteristic processing module 01 is used for carrying out discretization and vectorization processing on the user characteristics to obtain a plurality of user discrete characteristic vectors of the user characteristics; inputting a plurality of user discrete feature vectors of the user features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the user features; vector accumulation is carried out on a plurality of user discrete feature vectors of the user features to obtain linear output vectors of the user features; vector splicing is carried out on the deep learning output vector and the linear output vector of the user characteristic, and an output total vector of the user characteristic is obtained;
the project feature processing module 02 is used for carrying out discretization and vectorization processing on project features to obtain a plurality of project discrete feature vectors of the project features; inputting a plurality of project discrete feature vectors of project features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the project features; vector accumulation is carried out on a plurality of project discrete feature vectors of the project features, so that linear output vectors of the project features are obtained; vector splicing is carried out on the deep learning output vector and the linear output vector of the project feature, and an output total vector of the project feature is obtained;
The calculation model modeling module 03 is used for multiplying the output total vector of the user features and the output total vector of the project features by dot products to obtain a calculation model of the matching degree between the user features and the project features;
the matching degree calculating module 04 is used for inputting the user characteristics of the recall target user and the item characteristics of the recall target item into the calculating model to obtain the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item;
and the recommendation system recall module 05 is used for carrying out recall operation according to the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item.
In one embodiment, the user feature processing module is specifically configured to:
vector accumulation is carried out on a plurality of user discrete feature vectors of the user features according to the following formula, so that linear output vectors of the user features are obtained:
wherein U is FM A linear output vector representing a user characteristic; u (U) i Representing an ith user discrete feature vector; m represents the number of discrete feature vectors of the user.
In one embodiment, the project feature processing module is specifically configured to:
vector accumulation is carried out on a plurality of project discrete feature vectors of the project features according to the following formula, so that a linear output vector of the project features is obtained:
Wherein I is FM Line representing item featuresA sexual output vector; i j Representing a j-th item discrete feature vector; n represents the number of discrete feature vectors of the item.
In one embodiment, the user feature processing module is specifically configured to:
vector splicing is carried out on the deep learning output vector and the linear output vector of the user characteristic according to the following formula, so that the total output vector of the user characteristic is obtained:
wherein U is FM A linear output vector representing a user characteristic; u (U) Deep A deep learning output vector representing a user feature; u represents the output total vector of the user features;representing vector concatenation operations.
In one embodiment, the project feature processing module is specifically configured to:
vector splicing is carried out on the deep learning output vector and the linear output vector of the project feature according to the following formula, so that the total output vector of the project feature is obtained:
wherein I is FM A linear output vector representing the item feature; i Deep A deep learning output vector representing project features; i respectively represents the output total vector of project characteristics;representing vector concatenation operations.
In one embodiment, the computing model modeling module of the matching degree between the user characteristics and the project characteristics is specifically configured to:
and multiplying the output total vector of the user features and the output total vector of the project features by dot products according to the following formula to obtain a calculation model of the matching degree between the user features and the project features:
Wherein Score represents a computational model of the degree of matching between user features and project features; u (U) i And I i Respectively representing an ith user discrete feature vector and a jth item discrete feature vector; m and n represent the number of user discrete feature vectors and project discrete feature vectors, respectively;wherein U is FM And I FM A linear output vector representing the user feature and a linear output vector representing the project feature, respectively, U Deep And I Deep The deep learning output vector and the deep learning output vector of the project feature respectively represent the user feature, and U and I represent the total output vector of the user feature and the total output vector of the project feature respectively, +.>Representing a vector concatenation operation; /> Wherein U is i And I j Respectively representing an ith user discrete feature vector and a jth item discrete feature vector.
In one embodiment, as shown in fig. 4, further comprising: model modification module 06 for:
correcting the matching degree calculation model of the user characteristics and the project characteristics according to the cross entropy loss function to obtain a corrected matching degree calculation model of the user characteristics and the project characteristics;
inputting user characteristics of a recall target user and item characteristics of a recall target item into the computing model, including:
And inputting the user characteristics of the recall target user and the item characteristics of the recall target item into a corrected calculation model of the matching degree between the user characteristics and the item characteristics.
Based on the same inventive concept, an embodiment of the present invention further provides a computer device, and fig. 5 is a schematic diagram of a computer device provided in the embodiment of the present invention, as shown in fig. 5, the computer device 50 includes a memory 501, a processor 502, and a computer program stored on the memory 503 and capable of running on the processor 502, where the processor 502 implements the method for recalling the recommendation system when executing the computer program.
Based on the same inventive concept, the embodiments of the present invention also provide a computer-readable storage medium storing a computer program for executing the above-mentioned recommendation system recall method.
In the embodiment of the invention, discretization and vectorization are carried out on the user characteristics to obtain a plurality of user discrete characteristic vectors of the user characteristics; inputting a plurality of user discrete feature vectors of the user features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the user features; vector accumulation is carried out on a plurality of user discrete feature vectors of the user features to obtain linear output vectors of the user features; vector splicing is carried out on the deep learning output vector and the linear output vector of the user characteristic, and an output total vector of the user characteristic is obtained; discretizing and vectorizing the project features to obtain a plurality of project discrete feature vectors of the project features; inputting a plurality of project discrete feature vectors of project features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the project features; vector accumulation is carried out on a plurality of project discrete feature vectors of the project features, so that linear output vectors of the project features are obtained; vector splicing is carried out on the deep learning output vector and the linear output vector of the project feature, and an output total vector of the project feature is obtained; multiplying the output total vector of the user features and the output total vector of the project features by dot products to obtain a calculation model of the matching degree between the user features and the project features; inputting the user characteristics of the recall target user and the item characteristics of the recall target item into the calculation model to obtain the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item; according to the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item, recall operation is carried out, so that the accurate calculation of the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item can be realized by establishing a calculation model of the matching degree between the user characteristics and the item characteristics, the calculation of linear discrete characteristic vectors can be realized by determining linear output vectors, the problem that the recall capability of a recommendation system is limited because the traditional deep learning model can only carry out vector form calculation in the prior art is avoided, the aim of carrying out explicit second-order intersection of each characteristic on the discrete characteristic vectors can be realized by linearly outputting the vectors, and the recall accuracy of the recommendation system is improved; meanwhile, through linear output vectors, the recall of the recommendation system can be enabled to have memory capacity based on explicit feature cross combination, and recall efficiency of the recommendation system is improved.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing description of the embodiments has been provided for the purpose of illustrating the general principles of the invention, and is not meant to limit the scope of the invention, but to limit the invention to the particular embodiments, and any modifications, equivalents, improvements, etc. that fall within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (16)
1. A method of recommending system recall, comprising:
discretizing and vectorizing the user features to obtain a plurality of user discrete feature vectors of the user features; inputting a plurality of user discrete feature vectors of the user features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the user features; vector accumulation is carried out on a plurality of user discrete feature vectors of the user features to obtain linear output vectors of the user features; vector splicing is carried out on the deep learning output vector and the linear output vector of the user characteristic, and an output total vector of the user characteristic is obtained;
discretizing and vectorizing the project features to obtain a plurality of project discrete feature vectors of the project features; inputting a plurality of project discrete feature vectors of project features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the project features; vector accumulation is carried out on a plurality of project discrete feature vectors of the project features, so that linear output vectors of the project features are obtained; vector splicing is carried out on the deep learning output vector and the linear output vector of the project feature, and an output total vector of the project feature is obtained;
multiplying the output total vector of the user features and the output total vector of the project features by dot products to obtain a calculation model of the matching degree between the user features and the project features;
Inputting the user characteristics of the recall target user and the item characteristics of the recall target item into the calculation model to obtain the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item;
and carrying out recall operation according to the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item.
2. The method of claim 1, wherein vector-accumulating the plurality of user discrete feature vectors of the user feature to obtain the linear output vector of the user feature is performed according to the following formula:
wherein U is FM A linear output vector representing a user characteristic; u (U) i Representing an ith user discrete feature vector; m represents the number of discrete feature vectors of the user.
3. The method of claim 1, wherein vector-accumulating the plurality of item discrete feature vectors of the item feature to obtain a linear output vector of the item feature is performed according to the following formula:
wherein I is FM A linear output vector representing the item feature; i j Representing a j-th item discrete feature vector; n represents the number of discrete feature vectors of the item.
4. The method of claim 1, wherein the vector concatenation is performed on the deep learning output vector and the linear output vector of the user feature according to the following formula to obtain an output total vector of the user feature:
Wherein U is FM A linear output vector representing a user characteristic; u (U) Deep A deep learning output vector representing a user feature; u represents the output total vector of the user features;representing vector concatenation operations.
5. The method of claim 1, wherein the deep learning output vector and the linear output vector of the project feature are vector-stitched according to the following formula to obtain an output total vector of the project feature:
wherein I is FM A linear output vector representing the item feature; i Deep A deep learning output vector representing project features; i respectively represents the output total vector of project characteristics;representing vector concatenation operations.
6. The method of claim 1, wherein the dot product is multiplied by the total output vector of the user feature and the total output vector of the item feature according to the following formula to obtain a calculation model of the degree of matching between the user feature and the item feature:
wherein Score represents a computational model of the degree of matching between user features and project features; u (U) i And I i Respectively representing an ith user discrete feature vector and a jth item discrete feature vector; m and n represent the number of user discrete feature vectors and project discrete feature vectors, respectively;wherein U is FM And I FM A linear output vector representing the user feature and a linear output vector representing the project feature, respectively, U Deep And I Deep The deep learning output vector and the deep learning output vector of the project feature respectively represent the user feature, and U and I represent the total output vector of the user feature and the total output vector of the project feature respectively, +.>Representing a vector concatenation operation; /> Wherein U is i And I j Respectively representing an ith user discrete feature vector and a jth item discrete feature vector.
7. The method as recited in claim 1, further comprising:
correcting the matching degree calculation model of the user characteristics and the project characteristics according to the cross entropy loss function to obtain a corrected matching degree calculation model of the user characteristics and the project characteristics;
inputting user features of a recall target user and item features of a recall target item into the computing model, comprising:
and inputting the user characteristics of the recall target user and the item characteristics of the recall target item into a corrected calculation model of the matching degree between the user characteristics and the item characteristics.
8. An apparatus for recommending system recall, comprising:
the user feature processing module is used for carrying out discretization and vectorization processing on the user features to obtain a plurality of user discrete feature vectors of the user features; inputting a plurality of user discrete feature vectors of the user features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the user features; vector accumulation is carried out on a plurality of user discrete feature vectors of the user features to obtain linear output vectors of the user features; vector splicing is carried out on the deep learning output vector and the linear output vector of the user characteristic, and an output total vector of the user characteristic is obtained;
The project feature processing module is used for carrying out discretization and vectorization processing on project features to obtain a plurality of project discrete feature vectors of the project features; inputting a plurality of project discrete feature vectors of project features into a multi-layer fully-connected neural network to obtain a deep learning output vector of the project features; vector accumulation is carried out on a plurality of project discrete feature vectors of the project features, so that linear output vectors of the project features are obtained; vector splicing is carried out on the deep learning output vector and the linear output vector of the project feature, and an output total vector of the project feature is obtained;
the computing model modeling module is used for multiplying the output total vector of the user features and the output total vector of the project features by dot products to obtain a computing model of the matching degree between the user features and the project features;
the matching degree calculation module is used for inputting the user characteristics of the recall target user and the item characteristics of the recall target item into the calculation model to obtain the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item;
and the recommendation system recall module is used for carrying out recall operation according to the matching degree between the user characteristics of the recall target user and the item characteristics of the recall target item.
9. The apparatus of claim 8, wherein the user feature processing module is specifically configured to:
vector accumulation is carried out on a plurality of user discrete feature vectors of the user features according to the following formula, so that linear output vectors of the user features are obtained:
wherein U is FM A linear output vector representing a user characteristic; u (U) i Representing an ith user discrete feature vector; m represents the number of discrete feature vectors of the user.
10. The apparatus of claim 8, wherein the project feature processing module is specifically configured to:
vector accumulation is carried out on a plurality of project discrete feature vectors of the project features according to the following formula, so that a linear output vector of the project features is obtained:
wherein I is FM A linear output vector representing the item feature; i j Representing a j-th item discrete feature vector; n represents the number of discrete feature vectors of the itemAmount of the components.
11. The apparatus of claim 8, wherein the user feature processing module is specifically configured to:
vector splicing is carried out on the deep learning output vector and the linear output vector of the user characteristic according to the following formula, so that the total output vector of the user characteristic is obtained:
wherein U is FM A linear output vector representing a user characteristic; u (U) Deep A deep learning output vector representing a user feature; u represents the output total vector of the user features; Representing vector concatenation operations.
12. The apparatus of claim 8, wherein the project feature processing module is specifically configured to:
vector splicing is carried out on the deep learning output vector and the linear output vector of the project feature according to the following formula, so that the total output vector of the project feature is obtained:
wherein I is FM A linear output vector representing the item feature; i Deep A deep learning output vector representing project features; i respectively represents the output total vector of project characteristics;representing vector concatenation operations.
13. The apparatus of claim 8, wherein the computational model modeling module for matching the user features to the item features is specifically configured to:
and multiplying the output total vector of the user features and the output total vector of the project features by dot products according to the following formula to obtain a calculation model of the matching degree between the user features and the project features:
wherein Score represents a computational model of the degree of matching between user features and project features; u (U) i And I i Respectively representing an ith user discrete feature vector and a jth item discrete feature vector; m and n represent the number of user discrete feature vectors and project discrete feature vectors, respectively;wherein U is FM And I FM A linear output vector representing the user feature and a linear output vector representing the project feature, respectively, U Deep And I Deep The deep learning output vector and the deep learning output vector of the project feature respectively represent the user feature, and U and I represent the total output vector of the user feature and the total output vector of the project feature respectively, +.>Representing a vector concatenation operation; /> Wherein U is i And I j Respectively representing an ith user discrete feature vector and a jth item discrete feature vector.
14. The apparatus as recited in claim 8, further comprising: the model correction module is used for:
correcting the matching degree calculation model of the user characteristics and the project characteristics according to the cross entropy loss function to obtain a corrected matching degree calculation model of the user characteristics and the project characteristics;
inputting user features of a recall target user and item features of a recall target item into the computing model, comprising:
and inputting the user characteristics of the recall target user and the item characteristics of the recall target item into a corrected calculation model of the matching degree between the user characteristics and the item characteristics.
15. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method of any of claims 1 to 7 when executing the computer program.
16. A computer readable storage medium, characterized in that the computer readable storage medium stores a computer program for executing the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111088806.7A CN113742594B (en) | 2021-09-16 | 2021-09-16 | Recommendation system recall method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111088806.7A CN113742594B (en) | 2021-09-16 | 2021-09-16 | Recommendation system recall method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113742594A CN113742594A (en) | 2021-12-03 |
| CN113742594B true CN113742594B (en) | 2024-02-27 |
Family
ID=78739403
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111088806.7A Active CN113742594B (en) | 2021-09-16 | 2021-09-16 | Recommendation system recall method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113742594B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110866191A (en) * | 2019-11-21 | 2020-03-06 | 苏州朗动网络科技有限公司 | Recommendation recall method, apparatus and storage medium |
| CN111062775A (en) * | 2019-12-03 | 2020-04-24 | 中山大学 | Recommendation system recall method based on attention mechanism |
| CN111127145A (en) * | 2019-12-17 | 2020-05-08 | 武汉海云健康科技股份有限公司 | Sorting recommendation method and system based on combination of catboost algorithm and deep learning |
| CN112541130A (en) * | 2020-12-07 | 2021-03-23 | 东北大学 | Deep cross feature fusion based recommendation method and device |
-
2021
- 2021-09-16 CN CN202111088806.7A patent/CN113742594B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110866191A (en) * | 2019-11-21 | 2020-03-06 | 苏州朗动网络科技有限公司 | Recommendation recall method, apparatus and storage medium |
| CN111062775A (en) * | 2019-12-03 | 2020-04-24 | 中山大学 | Recommendation system recall method based on attention mechanism |
| CN111127145A (en) * | 2019-12-17 | 2020-05-08 | 武汉海云健康科技股份有限公司 | Sorting recommendation method and system based on combination of catboost algorithm and deep learning |
| CN112541130A (en) * | 2020-12-07 | 2021-03-23 | 东北大学 | Deep cross feature fusion based recommendation method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113742594A (en) | 2021-12-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107688855B (en) | Hierarchical quantization method and device for complex neural network | |
| CN106815252B (en) | Searching method and device | |
| US20210125516A1 (en) | Answer training device, answer training method, answer generation device, answer generation method, and program | |
| CN111814967B (en) | Method, apparatus and storage medium for calculating inferential computation of neural network model | |
| CN113641819A (en) | Multi-task sparse sharing learning-based argument mining system and method | |
| US20210158136A1 (en) | Calculation scheme decision system, calculation scheme decision device, calculation scheme decision method, and storage medium | |
| CN112463989B (en) | Knowledge graph-based information acquisition method and system | |
| CN111291187A (en) | Emotion analysis method and device, electronic equipment and storage medium | |
| CN116775497B (en) | Database test case generation demand description coding method | |
| CN109271632B (en) | Supervised word vector learning method | |
| Xie et al. | Overflow aware quantization: Accelerating neural network inference by low-bit multiply-accumulate operations | |
| CN114065769B (en) | Method, device, equipment and medium for training emotion reason pair extraction model | |
| CN117829149B (en) | Language model hybrid training method and device, electronic equipment and storage medium | |
| CN117574915A (en) | Public data platform based on multiparty data sources and data analysis method thereof | |
| US20210374545A1 (en) | Method and apparatus of increasing knowledge based on uncertainty in neural networks | |
| CN111460302B (en) | Data processing method, device, electronic equipment and computer readable storage medium | |
| CN113742594B (en) | Recommendation system recall method and device | |
| CN114911894A (en) | Information retrieval method, related system and storage medium | |
| Arasi et al. | Automated Image Captioning Using Sparrow Search Algorithm With Improved Deep Learning Model | |
| CN111563783B (en) | Article recommendation method and device | |
| CN111401568A (en) | Efficient machine learning algorithm, device and readable storage medium based on normal equation | |
| CN115880486B (en) | Target detection network distillation method and device, electronic equipment and storage medium | |
| CN114492395B (en) | Knowledge graph-oriented joint entity disambiguation method and system | |
| CN104298704A (en) | Method and system for pushing text on blog | |
| CN116049413B (en) | User viewpoint and position acquisition method based on event evolution |
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 |