CN110737834B - Recommendation method and device for business objects, storage medium and computer equipment - Google Patents
Recommendation method and device for business objects, storage medium and computer equipment Download PDFInfo
- Publication number
- CN110737834B CN110737834B CN201910973183.8A CN201910973183A CN110737834B CN 110737834 B CN110737834 B CN 110737834B CN 201910973183 A CN201910973183 A CN 201910973183A CN 110737834 B CN110737834 B CN 110737834B
- Authority
- CN
- China
- Prior art keywords
- text
- business
- service
- objects
- target
- 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 81
- 238000003860 storage Methods 0.000 title claims abstract description 15
- 239000013598 vector Substances 0.000 claims abstract description 298
- 230000008569 process Effects 0.000 claims abstract description 17
- 230000001960 triggered effect Effects 0.000 claims abstract description 14
- 238000012549 training Methods 0.000 claims description 24
- 230000006399 behavior Effects 0.000 claims description 23
- 238000012163 sequencing technique Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 8
- 238000013507 mapping Methods 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 description 38
- 230000000875 corresponding effect Effects 0.000 description 20
- 238000011176 pooling Methods 0.000 description 20
- 238000010586 diagram Methods 0.000 description 17
- 239000000284 extract Substances 0.000 description 15
- 230000009467 reduction Effects 0.000 description 8
- 230000006870 function Effects 0.000 description 6
- 238000006243 chemical reaction Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 235000021419 vinegar Nutrition 0.000 description 5
- 239000000052 vinegar Substances 0.000 description 5
- 230000004913 activation Effects 0.000 description 4
- 238000003062 neural network model Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000006855 networking Effects 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 235000013409 condiments Nutrition 0.000 description 2
- 230000002596 correlated effect Effects 0.000 description 2
- 235000013305 food Nutrition 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 235000012149 noodles Nutrition 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 230000003796 beauty Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 235000013361 beverage Nutrition 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 238000011524 similarity measure Methods 0.000 description 1
- 235000011888 snacks Nutrition 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000012360 testing method Methods 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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application relates to a recommendation method, a recommendation device, a storage medium and computer equipment of a business object, wherein the method comprises the following steps: acquiring a text vector of a triggered target business object; acquiring text vectors of other business objects except the target business object, wherein the text vectors of the target business object and the other business objects are determined according to a trained object class identification model; and determining a service object to be recommended according to the text vector of the target service object and the text vectors of the other service objects, and outputting the service object to be recommended. When the similar business objects are recommended, the complex operation process is avoided.
Description
Technical Field
The present invention relates to the field of computer technologies, and in particular, to a method and apparatus for recommending a service object, a storage medium, and a computer device.
Background
In conventional internet technology, a web site publishes a business object for a user logging into the web site to search, browse, and perform subsequent processing operations with respect to the particular business object. Taking a shopping website as an example, a business object is a commodity released by a seller, and a user logging in the shopping website searches for and browses the commodity, and collects, purchases and the like aiming at the commodity of interest; taking an information website as an example, a business object is news released by a publisher, and a user logging in the information website searches, browses news, and reviews, forwards and the like aiming at the news of interest.
When a user searches or browses a business object, the website recommends some business objects which the user may be interested in to the user, so that the user can quickly find the business objects which the user needs from the recommended business objects, for example, a shopping website can push commodities similar to commodities searched or browsed by the user, an information website can push news similar to news searched or browsed by the user, a resource website such as videos, music and books can push resources similar to resources searched or browsed by the user, a social account similar to a social account searched or browsed by the user, feed stream information similar to Feed stream information searched or browsed by the user and the like. However, in the traditional recommendation mode of similar service objects, word segmentation operation or word frequency statistics needs to be performed on text information of the service objects, so that the operation process is complicated.
Disclosure of Invention
Based on this, it is necessary to provide a service object recommendation method, apparatus, storage medium and computer device, aiming at the problem that the operation process of the conventional service object recommendation method is complicated.
A method of recommending business objects, the method comprising:
Acquiring a text vector of a triggered target business object;
acquiring text vectors of other business objects except the target business object, wherein the text vectors of the business objects are determined according to the trained object class identification model, and the business objects comprise the target business object and the other business objects;
and determining a service object to be recommended according to the text vector of the target service object and the text vectors of the other service objects, and outputting the service object to be recommended.
A recommendation device for a business object, the device comprising:
the acquisition module is used for acquiring the text vector of the triggered target business object;
the obtaining module is further configured to obtain text vectors of other service objects except the target service object, where the text vectors of the service objects are determined according to the trained object class identification model, and the service objects include the target service object and the other service objects;
and the determining module is used for determining the service object to be recommended according to the text vector of the target service object and the text vectors of the other service objects and outputting the service object to be recommended.
A storage medium having stored thereon computer executable instructions which, when executed by a processor, cause the processor to perform the steps of a business object recommendation method.
A computer device comprising a memory and a processor, the memory having stored therein computer readable instructions which, when executed by the processor, cause the processor to perform the steps of a business object recommendation method.
The method, the device, the storage medium and the computer equipment for recommending the service object acquire the text vector of the triggered target service object, acquire the text vectors of other service objects except the target service object, determine the service object to be recommended according to the text vector of the target service object and the text vectors of other service objects, and output the service object to be recommended, wherein the text vectors of the target service object and the other service objects are determined according to the trained object class identification model. According to the service object recommending method, the text vector of the service object is determined according to the object type recognition model, so that a complicated operation process is avoided, meanwhile, the object type recognition model considers the type of the service object and the semantic meaning of text information of the service object, similar service objects can be recommended in the dimension of the type, the influence of the word ambiguity on the similarity between the service objects is avoided, and the conversion rate of recommended information is improved.
Drawings
FIG. 1 is an internal block diagram of a terminal for implementing a business object recommendation method in one embodiment;
FIG. 2 is a flow diagram of a business object recommendation method in one embodiment;
FIG. 3 is a block diagram of the structure of an object class identification model in one embodiment;
FIG. 4 is a block diagram of an object class identification model in another embodiment;
FIG. 5 is a block diagram of an object class identification model in yet another embodiment;
FIG. 6 is a schematic diagram of a recommendation result of a business object in one embodiment;
FIG. 7 is a schematic diagram illustrating the operation of an embedding layer in one embodiment;
FIG. 8 is a schematic diagram of the functioning of a convolution layer in one embodiment;
FIG. 9 is a schematic diagram of the role of the pooling layer in one embodiment;
FIG. 10 is a flowchart of another embodiment of a method for recommending business objects;
FIG. 11 is a flowchart of a method for recommending business objects according to another embodiment;
FIG. 12 is a block diagram of a business object recommender in one embodiment;
FIG. 13 is a block diagram illustrating a business object recommendation device in another embodiment;
fig. 14 is an internal structural diagram of a computer device in one embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.
Fig. 1 is a schematic diagram of an internal structure of a terminal in one embodiment. As shown in fig. 1, the terminal includes a processor, a nonvolatile storage medium, an internal memory and a network interface, a display screen, and an input device connected by a system bus. The non-volatile storage medium of the terminal stores an operating system and may further store computer readable instructions that, when executed by the processor, cause the processor to implement a business object recommendation method. The processor is configured to provide computing and control capabilities to support the operation of the entire terminal. The internal memory may also have stored therein computer readable instructions that, when executed by the processor, cause the processor to perform a business object recommendation method. The network interface is used for network communication with a server or other terminal. The display screen of the terminal can be a liquid crystal display screen or an electronic ink display screen, etc. The input device can be a touch layer covered on the display screen, can also be a key, a track ball or a touch pad arranged on the terminal shell, and can also be an external keyboard, a touch pad or a mouse, etc.
It will be appreciated by those skilled in the art that the structure shown in fig. 1 is merely a block diagram of a portion of the structure associated with the present application and is not limiting of the terminal to which the present application is applied, and that a particular terminal may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.
As shown in FIG. 2, in one embodiment, a business object recommendation method is provided.
Referring to fig. 2, the embodiment is mainly illustrated by the application of the method to the terminal in fig. 1, and the recommendation method of the service object specifically includes the following steps:
s202, acquiring a text vector of the triggered target business object.
The service object refers to an object which can be used as an operation target when user operation behaviors are executed. The user operation behavior refers to interaction behavior between the user and the terminal, for example, a selection operation such as clicking, sliding, etc. performed by the user on a display interface of the terminal, a search operation performed by the user by inputting keywords, titles, links, passwords, etc. in a search area of the terminal, etc. The business objects can be commodities on the shopping website, news released by the information website, videos released by the video website, music on the music website, books uploaded by the book website, account numbers of the social networking website, feed streams and the like. The target service object refers to an object selected as an operation target by the user operation behavior. For example, the user clicks on a commodity in the shopping website, and the commodity is the target business object.
The text vector refers to text information representing the business object in a vector mode, and the vector used for representing can be a low-dimensional vector. Text information is information related to a business object, such as a title, a keyword, a brief introduction, a abstract, etc. of the business object. Text vectors may characterize how far or near a relationship between business objects is, the closer the relationship between business objects is, the smaller the vector distance between their corresponding text vectors. Taking a business object as an example, the vector distance between the text vector of the skirt and the text vector of the trousers is closer than the vector distance between the text vector of the skirt and the text vector of the necklace.
Specifically, the text vector of the target business object is determined according to the object category recognition model obtained through training. The object type recognition model can recognize the type of the service object, in the recognition process, the object type recognition model can generate text vectors of text information of the service object, and then the type of the service object is determined according to the text vectors of the service object, so that the text vectors of the service object can reflect the relationship between the service objects. And testing the object type recognition model obtained through training, wherein the classification accuracy of the object type recognition model on the text information of the service object can reach 92%, and the text vector of the service object determined according to the object type recognition model is explained to be capable of accurately representing the semantics of the text information of the service object.
The category of the business object can be characterized by the characteristics of the category, the source and the like of the business object.
The categories of business objects are categories that are divided into categories based on the nature of the business object itself. Taking a business object as an example, the categories may include: snack foods, drinks, beverages, grain and oil, staple foods, clothing, personal care, household kitchen ware, household cleaning, mother and infant toys, household digital appliances, medical health care, beauty and skin care, pet life and the like; taking business objects as news, video, music, books, account numbers of social networking sites or Feed streams as examples, the categories can be divided by the domain to which they belong, for example: science and technology, finance, sports, entertainment, automobiles, real estate, fashion, education, travel, games, and the like.
The source of a business object refers to information related to the generation of the business object. Taking a business object as an example, sources may include: production place, production manufacturer, brand, etc.; taking business objects as news, video, music, books, accounts of social networking sites or Feed streams as examples, categories may include: publishers, authors, etc.
In particular, the object class identification model may be a neural network model. As shown in fig. 3, the object class identification model may include an input layer, an hidden layer, and an output layer, with the links between the input layer and the hidden layer being established sequentially, and finally the links between the hidden layer and the output layer being established.
In one embodiment, as shown in FIG. 4, the object class identification model may include an input layer, an embedded layer, an implied layer, and a classifier. The input layer is used for receiving the sparse matrix of the input business object and sending the sparse matrix of the business object to the embedded layer; the embedded layer performs dimension reduction operation on the sparse matrix of the input business object to obtain a dense matrix, and sends the dense matrix to the hidden layer; the hidden layer extracts distinguishing features based on the input dense matrix and sends the extracted features to the classifier; the classifier calculates the probability that the business object belongs to each category, thereby achieving the purpose of classification.
In one embodiment, as shown in FIG. 5, the hidden layers may include a convolutional layer, a pooled layer, and a fully-connected layer. The convolution layer initially extracts features on the input dense matrix to obtain a convolution feature vector; the pooling layer performs dimension reduction operation on the convolution feature vector output by the convolution layer, and further extracts features to obtain a pooling feature vector; and the full connection layer combines or samples the feature vectors obtained by the processing of the convolution layer and the pooling layer to extract distinguishing features.
The object type recognition model can be obtained through text information of a sample service object and type training to which the sample service object belongs, and the sample service object is the service object for training the object type recognition model. The text information of the sample business object is marked by utilizing the category to which the sample business object belongs, the text information of the marked sample business object is input into an object category identification model, and the object category identification model is trained in a back propagation mode, so that the identification result of the object category identification model approaches to the mark of the sample business object, and the object category identification model learns the mapping between the text information of the sample business object and the category to which the sample business object belongs.
In one embodiment, the manner of obtaining the text vector of the target business object corresponding to the user operation behavior may be: and acquiring a matching relation between the pre-stored business object and the text vector, and searching the text vector of the target business object corresponding to the user operation behavior according to the matching relation. Wherein the matching relationship is established based on the business object and the text vector of the business object determined by the object class recognition model. The text vector of the service object is determined in advance through the object type recognition model obtained through training, and the service object and the text vector of the service object are stored in a correlated mode, so that the text vector of the target service object is directly called according to the target service object when the service object is applied, and the calculation process is reduced.
In another embodiment, the manner of obtaining the text vector of the target business object corresponding to the user operation behavior may be: and acquiring text information of the target business object corresponding to the user operation behavior, and inputting the text information of the target business object into an object class identification model to obtain a text vector of the target business object. Specifically, inputting text information of the target service object into an object class identification model to obtain a text vector of the target service object output by a preset layer in an implicit layer of the object class identification model. The preset layer may be any layer in the hidden layers. Optionally, the preset layer is the last full-connection layer.
S204, obtaining text vectors of other business objects except the target business object, wherein the text vectors of the business objects are determined according to the trained object class identification model, and the business objects comprise the target business object and the other business objects.
The other business objects refer to objects, which are not selected as operation targets by the user operation behaviors, such as goods, which are not clicked by the user, in a shopping website, and the goods are the other business objects.
Specifically, text vectors of other business objects are determined according to the trained object class recognition model.
In one embodiment, the manner of obtaining the text vector of the other business object corresponding to the user operation behavior may be: and acquiring a matching relation between the pre-stored business object and the text vector, and searching the text vectors of other business objects corresponding to the user operation behaviors according to the matching relation. Wherein the matching relationship is established based on the business object and the text vector of the business object determined by the object class recognition model. The text vector of the service object is determined in advance through the object type recognition model obtained through training, and the service object and the text vector of the service object are stored in a correlated mode, so that the text vector of other service objects is directly called according to the other service objects when the service object is applied, and the calculation process is reduced.
In another embodiment, the manner of obtaining the text vector of the other business object corresponding to the user operation behavior may be: text information of other business objects corresponding to the user operation behaviors is obtained, the text information of the other business objects is input into the object class identification model, and text vectors of the other business objects are obtained. Specifically, text information of other business objects is input into an object type recognition model to obtain text vectors of the other business objects output by a preset layer in an implicit layer of the object type recognition model. The preset layer may be any layer in the hidden layers. Optionally, the preset layer is the last full-connection layer.
S206, determining the business object to be recommended according to the text vector of the target business object and the text vectors of the other business objects, and outputting the business object to be recommended.
Specifically, the manner of determining the service object to be recommended according to the text vector of the target service object and the text vectors of other service objects may be: and obtaining the similarity between the text vector of the target service object and the text vectors of other service objects, and determining the service object to be recommended in the other service objects according to the similarity.
In one embodiment, the similarity between the text vector of the target business object and the text vector of the other business object may be determined by the vector distance between the text vector of the target business object and the text vector of the other business object.
Wherein the vector distance is inversely proportional to the similarity, i.e. the smaller the vector distance between the text vector of the target business object and the text vector of the other business object, the greater the similarity between the text vector of the target business object and the text vector of the other business object. In particular, the vector distance between the text vector of the target business object and the text vector of the other business objects may be characterized by euclidean distance, cosine similarity, and the like.
In one embodiment, the manner of determining the service object to be recommended among other service objects according to the similarity may be: and sequencing other business objects by using the similarity, and determining the business objects to be recommended according to the sequencing result.
Specifically, after the similarity between the text vector of the target business object and the text vectors of other business objects is obtained, the text vectors are ranked according to the similarity from large to small or the text vectors are ranked according to the similarity from small to large. And selecting other business objects with preset quantity according to the sequence from the high similarity to the low similarity according to the sequencing result, and taking the selected other business objects as business objects to be recommended.
As shown in fig. 6, taking a business object as an example, when a user clicks on a target commodity "Shanxi mature vinegar", text vectors of the target commodity and text vectors of other commodities are respectively determined through an object type recognition model, similarity between the text vectors of the target commodity and the text vectors of the other commodities is obtained, the other commodities are ranked by using the similarity, and ten other commodities are selected and recommended to the user according to the sequence from high similarity to low similarity. As can be seen from fig. 6, the commodity recommended to the user is not limited to "vinegar", but is recommended in a larger category, namely "condiment", so that the service object is recommended in the dimension of the category, and the recommended service object has relevance with the target object.
According to the service object recommending method, text vectors of triggered target service objects are obtained, text vectors of other service objects except the target service objects are obtained, the service objects to be recommended are determined according to the text vectors of the target service objects and the text vectors of other service objects, and the service objects to be recommended are output, wherein the text vectors of the target service objects and the text vectors of other service objects are determined according to the trained object type recognition model. According to the service object recommending method, the text vector of the service object is determined according to the object type recognition model, so that a complicated operation process is avoided, meanwhile, the object type recognition model considers the type of the service object and the semantic meaning of text information of the service object, similar service objects can be recommended in the dimension of the type, the influence of the word ambiguity on the similarity between the service objects is avoided, and the conversion rate of recommended information is improved.
In one embodiment, the obtaining the text vector of the target business object corresponding to the user operation behavior includes: acquiring a pre-stored matching relation between a business object and a text vector, wherein the matching relation is established according to the business object and the text vector of the business object determined by the object type recognition model; and searching the text vector of the target business object corresponding to the user operation behavior according to the matching relation.
Specifically, the text vector of the service object is determined in advance through the object type recognition model obtained through training, and the service object and the text vector of the service object are stored in an associated mode, so that the text vector of the target service object is directly called according to the target service object when the service object is applied, and complex operation processes such as word segmentation operation and word frequency statistics are avoided.
According to the service object recommendation method provided by the embodiment, the text vector of the target service object is searched through the matching relation according to the target service object, so that a complicated operation process is avoided.
In one embodiment, the obtaining the text vector of the target business object corresponding to the user operation behavior includes: acquiring text information of a target business object corresponding to the user operation behavior; and inputting the text information of the target business object into the object category identification model to obtain the text vector of the target business object.
Specifically, an object type recognition model is obtained through pre-training, text information of a target service object is input into the object type recognition model, and then a text vector of the target service object can be obtained.
In particular, the object class identification model may be a neural network model. As shown in fig. 3, the object class identification model may include an input layer, an hidden layer, and an output layer, with the links between the input layer and the hidden layer being established sequentially, and finally the links between the hidden layer and the output layer being established.
In one embodiment, as shown in FIG. 4, the object class identification model may include an input layer, an embedded layer, an implied layer, and a classifier. The input layer is used for receiving the sparse matrix of the input business object and sending the sparse matrix of the business object to the embedded layer; the embedded layer performs dimension reduction operation on the sparse matrix of the input business object to obtain a dense matrix, and sends the dense matrix to the hidden layer; the hidden layer extracts distinguishing features based on the input dense matrix and sends the extracted features to the classifier; the classifier calculates the probability that the business object belongs to each category, thereby achieving the purpose of classification.
In one embodiment, as shown in FIG. 5, the hidden layers may include a convolutional layer, a pooled layer, and a fully-connected layer. The convolution layer initially extracts features on the input dense matrix to obtain a convolution feature vector; the pooling layer performs dimension reduction operation on the convolution feature vector output by the convolution layer, and further extracts features to obtain a pooling feature vector; and the full connection layer combines or samples the feature vectors obtained by the processing of the convolution layer and the pooling layer to extract distinguishing features.
Specifically, inputting text information of the target service object into an object class identification model to obtain a text vector of the target service object output by a preset layer in an implicit layer of the object class identification model. The preset layer may be any layer in the hidden layers. Optionally, the preset layer is the last full-connection layer.
According to the service object recommending method, text information of the target service object is input into the object type recognition model, so that text vectors of the target service object can be obtained, and complicated operation processes are avoided.
In one embodiment, the object class identification model includes an implicit layer; inputting the text information of the target business object into the object category recognition model to obtain a text vector of the target business object, wherein the text vector comprises the following components: and inputting the text information of the target business object into the object category identification model to obtain the text vector of the target business object output by a preset layer in the hidden layer.
Before entering the text information of the business object into the object class recognition model, the text information of the business object needs to be preprocessed, i.e. a numerical representation of the text information of the business object is obtained, so as to convert the text information of the business object into a language understandable by a computer. The numerical representation of the textual information of the business object may be determined by a pre-constructed mapping table.
Before training the object class identification model, collecting text information of the sample business objects, unpacking each word in the text information of all the sample business objects into a set, and executing a deduplication operation on repeated words in the set. And respectively giving an identity identifier for each word in the set, and storing each word and the corresponding identity identifier in an associated manner to obtain a mapping table between the words and the identity identifiers. In one embodiment, the identity may be an integer value, such as 0,1, 2, etc. For example, there are seven words in the set-face, bar, convenience, face, woman, dress ", then the mapping table may be { face: 0, bar: 1, the following formula: 2, the following steps: 3, surface: 4, female: 5, loading: 6}.
The numerical representation of the text information of the business object can be converted into a sparse matrix by an One Hot coding mode. The sparse matrix may be in the form of a×b, where a is the total number of words in the set, B is the dimension of the vector corresponding to the word, and the dimension of the vector may be set according to practical applications, such as 32, 64, 128, 256, etc. In the sparse matrix, each row of vectors corresponds to a word of the text information of the business object. The length of each row of vectors is the total number of words in the set, only one 1 exists in each row of vectors, and the other positions which are all 0 and 1 correspond to the identity of the word in the mapping table. For example, the sparse matrix corresponding to "noodles" is noodles [1 0 0 0 0 0 0], bars [0 1 0 0 0 0 0]. In the sparse matrix, each word is independent, no correlation exists between semantics, and the sparse matrix is too long, so that resources are excessively occupied.
In particular, the object class identification model may be a neural network model. As shown in fig. 3, the object class identification model may include an input layer, an hidden layer, and an output layer, with the links between the input layer and the hidden layer being established sequentially, and finally the links between the hidden layer and the output layer being established.
In one embodiment, as shown in FIG. 4, the object class identification model may include an input layer, an embedded layer, an hidden layer, and a classifier, where the classifier implements the output function.
Specifically, the input layer is configured to receive a sparse matrix of the input business object, and send the sparse matrix of the business object to the embedding layer.
The embedding layer performs dimension reduction operation on the sparse matrix of the input business object to obtain a dense matrix, and sends the dense matrix to the hidden layer. As shown in fig. 7, the embedded layer corresponds to a simple neural network model, and the basic idea is to map the vector of the sparse matrix corresponding to each word into a short vector with a fixed length, so as to realize dimension reduction. Moreover, the embedded layer trains based on the same characteristics among the words, so that vectors among words with similar semantics have relevance in a dense matrix. Specifically, the embedding layer maps the vector of the sparse matrix corresponding to each word of the text information of the service object into a short vector with a fixed length, and maps all words in the text information of the service object into a dense matrix.
The hidden layer extracts the characteristic with distinguishing property based on the input dense matrix and sends the extracted characteristic to the classifier. An activation function is selected for each hidden layer node, the activation function being a function further enhanced after summing the inputs of the respective paths, optionally the activation function being a ReLU, tanh or Sigmoid.
The classifier calculates the probability that the business object belongs to each category, thereby achieving the purpose of classification. Alternatively, the classifier may be a Softmax activation function, an SVM classifier, or the like.
In one embodiment, the hidden layers may include a convolutional layer, a pooled layer, and a fully-connected layer. The number of convolution layers, pooling layers and full-connection layers may be set according to practical applications, and in one embodiment, as shown in fig. 5, one convolution layer, one pooling layer and two full-connection layers are selected.
Specifically, the convolution layer initially extracts features on an input dense matrix to obtain a convolution feature vector; the pooling layer performs dimension reduction operation on the convolution feature vector output by the convolution layer, and further extracts features to obtain a pooling feature vector; and the full connection layer combines or samples the feature vectors obtained by the processing of the convolution layer and the pooling layer to extract distinguishing features.
Specifically, inputting text information of the target service object into an object class identification model to obtain a text vector of the target service object output by a preset layer in an implicit layer of the object class identification model. The preset layer may be any layer in the hidden layers. Optionally, the preset layer is the last full-connection layer.
According to the service object recommendation method provided by the embodiment, the object type recognition model extracts the characteristics of the target service object layer by layer, so that the text vector of the target service object obtained through the object type recognition model fully considers the type of the service object and the semantics of the text information of the service object.
In one embodiment, the hidden layer comprises a fully connected layer; inputting the text information of the target service object into the object category recognition model to obtain a text vector of the target service object output by a preset layer in the hidden layer, wherein the text vector comprises the following components: and inputting the text information of the target business object into the object category identification model to obtain the text vector of the target business object output by the last full-connection layer in the hidden layers.
The convolution layer initially extracts features on the input dense matrix to obtain a convolution feature vector. As shown in fig. 8, the convolution layer is extracted by: and sliding the feature detectors in a small area and a small area on the input dense matrix, and calculating point multiplication by using the feature detectors to obtain a convolution feature vector.
The feature detector is a matrix vector of m×n, and the values of m and n can be set according to practical applications. For the same dense matrix, different m x n feature detectors will generate different convolution feature vectors. The more feature detectors used, the more features the convolutional layer extracts. The convolution eigenvector is controlled by three parameters: depth, stride, and zero-padding. The depth refers to the number of feature detectors used in convolution operation, the step length refers to the number of vectors passing through the feature detectors in the sliding direction when each sliding occurs, and the zero filling refers to filling with zero values at the edges of the input dense matrix, so that the edges of the input dense matrix can be filtered.
The pooling layer performs dimension reduction operation on the convolution feature vector output by the convolution layer, and further extracts features to obtain the pooled feature vector. Pooling may be achieved by: maximum pooling, average pooling, addition pooling, and the like. As shown in fig. 9, taking the maximum pooling as an example, the pooling layer makes the convolution feature vector smaller on the one hand, so as to simplify the computational complexity, and further performs feature extraction based on the convolution feature vector on the other hand.
The full connection layer combines or samples the feature vectors obtained by the processing of the convolution layer and the pooling layer, and extracts distinguishing features. The neurons between the layers of the fully connected layer are connected in a fully connected mode. In the object class identification model, there is at least one fully connected layer.
Specifically, inputting text information of a target service object into an object type recognition model to obtain a text vector output by a last full-connection layer of the object type recognition model, and taking the text vector as a text vector of the target service object.
According to the service object recommendation method provided by the embodiment, the object type recognition model extracts the characteristics of the target service object layer by layer, so that the text vector of the target service object obtained through the object type recognition model fully considers the type of the service object and the semantics of the text information of the service object.
In one embodiment, the determining the service object to be recommended according to the text vector of the target service object and the text vectors of the other service objects includes: obtaining the similarity between the text vector of the target service object and the text vectors of the other service objects; and determining the business object to be recommended from the other business objects according to the similarity.
In particular, the business object to be recommended may be determined among other business objects by a similarity between the text vector of the target business object and the text vector of the other business objects.
In one embodiment, the similarity between the text vector of the target business object and the text vector of the other business object may be characterized by a vector distance between the text vector of the target business object and the text vector of the other business object.
According to the service object recommending method, the service object to be recommended is determined in other service objects through the similarity between the text vector of the target service object and the text vectors of other service objects, and therefore accuracy of service object recommendation is improved.
In one embodiment, the obtaining the similarity between the text vector of the target business object and the text vector of the other business object includes: and obtaining vector distances between the text vectors of the target business object and the text vectors of the other business objects, and utilizing the vector distances to represent the similarity between the text vectors of the target business object and the text vectors of the other business objects.
Wherein the vector distance is inversely proportional to the similarity, i.e. the smaller the vector distance between the text vector of the target business object and the text vector of the other business object, the greater the similarity between the text vector of the target business object and the text vector of the other business object.
In particular, the vector distance between the text vector of the target business object and the text vector of the other business objects may be characterized by euclidean distance, cosine similarity, and the like. The Euclidean distance can be used for measuring the absolute distance between two individuals in the vector space, and the cosine similarity measures the difference between the two individuals by using the cosine value of the included angle of the two vectors in the vector space.
According to the service object recommendation method, the similarity between the text vector of the target service object and the text vector of other service objects is represented through the vector distance between the text vector of the target service object and the text vector of the other service objects, so that the accuracy of service object recommendation is improved.
In one embodiment, the determining the service object to be recommended among the other service objects according to the similarity includes: sorting the other business objects by using the similarity; and determining the business object to be recommended according to the sequencing result.
Specifically, after the similarity between the text vector of the target business object and the text vectors of other business objects is obtained, the text vectors are ranked according to the similarity from large to small or the text vectors are ranked according to the similarity from small to large.
According to the service object recommending method provided by the embodiment, other service objects are ordered by using the similarity, so that the recommended service objects are accurately selected.
In one embodiment, the determining the service object to be recommended according to the sorting result includes: and selecting a preset number of other service objects according to the sequence from the high similarity to the low similarity according to the sequencing result, and taking the selected other service objects as the service objects to be recommended.
The preset number may be set according to practical applications, for example, 5, 8, 10, etc.
As shown in fig. 6, taking a business object as an example, when the user clicks on the target commodity "Shanxi mature vinegar", 10 other commodities are selected and recommended to the user according to the order of the similarity from high to low.
According to the service object recommending method provided by the embodiment, the other service objects with the preset quantity are selected as the service objects to be recommended according to the sequence from the high similarity to the low similarity, so that the recommended service objects are accurately selected.
In one embodiment, the training mode of the object class identification model includes: acquiring text information of a sample service object and a category to which the sample service object belongs; and training the object category identification model by a back propagation mode according to the text information of the sample service object and the category to which the sample service object belongs.
Wherein, the sample business object refers to a business object used for training an object category identification model. The sample business object has a wider selection range, and takes the business object as an example, the sample business object can be the commodity of all shopping websites on the market.
The object type recognition model can be obtained through text information of the sample business object and type training of the sample business object. The text information of the sample business object is marked by the category to which the sample business object belongs, the text information of the marked sample business object is input into the object category identification model, so that the identification result of the object category identification model approaches to the mark of the sample business object, and the object category identification model learns the mapping between the text information of the sample business object and the category to which the sample business object belongs.
Specifically, training the object class recognition model by back propagation includes: when the probability that the sample business object belongs to each category is obtained, comparing the probability with the mark of the sample business object to obtain an error; calculating the gradient of the error according to the weight of the object class identification model by using a back propagation algorithm; and updating parameters of the object class identification model by using a gradient descent algorithm so as to minimize errors output by the object class identification model. Wherein the parameters of the object class identification model may include the weight of each join, the bias value of each node's own band, and the value of the matrix vector of the feature detector.
According to the service object recommendation method provided by the embodiment, the object class identification model is trained in a back propagation mode according to the text information of the sample service object and the class to which the sample service object belongs, so that the object class identification model can more accurately understand the class to which the service object belongs and the semantics of the text information of the service object.
As shown in fig. 10, in a specific embodiment, the recommendation method of the business object includes the following steps:
s1002, inputting text information of a triggered target service object into a trained object class identification model to obtain a text vector of the target service object output by a last full-connection layer of the object class identification model;
S1004, inputting text information of other business objects except the target business object into the object type recognition model to obtain text vectors of the other business objects output by the last full-connection layer of the object type recognition model;
s1006, obtaining vector distances between the text vectors of the target business object and the text vectors of the other business objects, and utilizing the vector distances to represent the similarity between the text vectors of the target business object and the text vectors of the other business objects;
s1008, sorting the other business objects by using the similarity, selecting a preset number of the other business objects according to the sequence from the high similarity to the low similarity, and taking the selected other business objects as business objects to be recommended;
s1010, outputting the business object to be recommended.
According to the service object recommending method, text vectors of triggered target service objects are obtained, text vectors of other service objects except the target service objects are obtained, the service objects to be recommended are determined according to the text vectors of the target service objects and the text vectors of other service objects, and the service objects to be recommended are output, wherein the text vectors of the target service objects and the text vectors of other service objects are determined according to the trained object type recognition model. According to the service object recommending method, the text vector of the service object is determined according to the object type recognition model, so that a complicated operation process is avoided, meanwhile, the object type recognition model considers the type of the service object and the semantic meaning of text information of the service object, similar service objects can be recommended in the dimension of the type, the influence of the word ambiguity on the similarity between the service objects is avoided, and the conversion rate of recommended information is improved.
As shown in fig. 11, in a specific embodiment, the recommendation method of the business object includes the following steps:
s1102, inputting the title of the triggered target commodity into a trained object type recognition model to obtain a text vector of the target commodity output by the last full-connection layer of the object type recognition model;
s1104, inputting titles of other commodities except the target commodity into the object type recognition model to obtain text vectors of the other commodities output by a last full-connection layer of the object type recognition model;
s1106, obtaining a vector distance between the text vector of the target commodity and the text vectors of other commodities, and utilizing the vector distance to represent the similarity between the text vector of the target commodity and the text vectors of the other commodities;
s1108, sorting the other commodities by using the similarity, selecting ten other commodities according to the sequence from the high similarity to the low similarity, and taking the ten selected other commodities as commodities to be recommended;
s1111, outputting the commodity to be recommended.
As shown in fig. 6, when the user clicks on the target commodity "Shanxi mature vinegar", text vectors of the target commodity and text vectors of other commodities are respectively determined through the object type recognition model, similarity between the text vectors of the target commodity and the text vectors of the other commodities is obtained, the other commodities are ranked by using the similarity, and ten other commodities are selected and recommended to the user according to the order of the similarity from large to small. As can be seen from fig. 6, the commodity recommended to the user is not limited to "vinegar", but is recommended in a larger category, namely "condiment", so that the service object is recommended in the dimension of the category, and the recommended service object has relevance with the target object.
According to the business object recommending method, text vectors of triggered target commodities are obtained, text vectors of other commodities except the target commodities are obtained, commodities to be recommended are determined according to the text vectors of the target commodities and the text vectors of the other commodities, and the commodities to be recommended are output, wherein the text vectors of the target commodities and the other commodities are determined according to the object type recognition model obtained through training. According to the recommendation method of the business object, the text vector of the commodity is determined according to the object type recognition model, so that a complicated operation process is avoided, meanwhile, the object type recognition model considers the type of the commodity and the semantic meaning of the title of the commodity, similar commodities can be recommended in the dimension of the type, the influence of the word ambiguity on the similarity between the commodities is avoided, and the conversion rate of recommendation information is improved.
Fig. 2, 10 and 11 are schematic flow diagrams of a method for recommending a business object in one embodiment. It should be understood that, although the steps in the flowcharts of fig. 2, 10, and 11 are shown in order as indicated by the arrows, these steps are not necessarily performed in order as indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in fig. 2, 10, and 11 may include multiple sub-steps or stages that are not necessarily performed at the same time, but may be performed at different times, nor does the order in which the sub-steps or stages are performed necessarily occur sequentially, but may be performed alternately or alternately with other steps or at least a portion of the sub-steps or stages of other steps.
As shown in fig. 12, in one embodiment, there is provided a recommendation apparatus 1200 for a business object, including: an acquisition module 1202 and a determination module 1204.
An obtaining module 1202, configured to obtain a text vector of a triggered target service object;
the obtaining module 1202 is further configured to obtain text vectors of other business objects except the target business object, where the text vectors of the business objects are determined according to the trained object class identification model, and the business objects include the target business object and the other business objects;
and the determining module 1204 is configured to determine a service object to be recommended according to the text vector of the target service object and the text vectors of the other service objects, and output the service object to be recommended.
The service object recommending apparatus 1200 acquires the text vector of the triggered target service object, acquires the text vectors of other service objects except the target service object, determines the service object to be recommended according to the text vector of the target service object and the text vectors of other service objects, and outputs the service object to be recommended, wherein the text vectors of the target service object and the other service objects are determined according to the trained object class identification model. The recommendation device 1200 of the business object determines the text vector of the business object according to the object category identification model, avoids a complicated operation process, and meanwhile, the object category identification model considers the category of the business object and the semantic of text information thereof, so that similar business objects can be recommended in the dimension of the category, the influence of the word ambiguity on the similarity between the business objects is avoided, and the conversion rate of the recommended information is improved.
In one embodiment, the acquiring module 1202 is further configured to: acquiring a pre-stored matching relation between a business object and a text vector, wherein the matching relation is established according to the business object and the text vector of the business object determined by the object type recognition model; and searching the text vector of the target business object corresponding to the user operation behavior according to the matching relation.
In one embodiment, the acquiring module 1202 is further configured to: acquiring text information of a target business object corresponding to the user operation behavior; and inputting the text information of the target business object into the object category identification model to obtain the text vector of the target business object.
In one embodiment, the object class identification model includes an implicit layer; the acquiring module 1202 is further configured to: and inputting the text information of the target business object into the object category identification model to obtain the text vector of the target business object output by a preset layer in the hidden layer.
In one embodiment, the hidden layer comprises a fully connected layer; the acquiring module 1202 is further configured to: and inputting the text information of the target business object into the object category identification model to obtain the text vector of the target business object output by the last full-connection layer in the hidden layers.
In one embodiment, the determining module 1204 is further configured to: obtaining the similarity between the text vector of the target service object and the text vectors of the other service objects; and determining the business object to be recommended from the other business objects according to the similarity.
In one embodiment, the determining module 1204 is further configured to: and obtaining vector distances between the text vectors of the target business object and the text vectors of the other business objects, and utilizing the vector distances to represent the similarity between the text vectors of the target business object and the text vectors of the other business objects.
In one embodiment, the determining module 1204 is further configured to: sorting the other business objects by using the similarity; and determining the business object to be recommended according to the sequencing result.
In one embodiment, the determining module 1204 is further configured to: and selecting a preset number of other service objects according to the sequence from the high similarity to the low similarity according to the sequencing result, and taking the selected other service objects as the service objects to be recommended.
In one embodiment, as shown in fig. 13, the recommendation device 1200 of the service object further includes: training module 1206, the obtaining module 1202 is further configured to: acquiring text information of a sample service object and a category to which the sample service object belongs; the training module 1206 is configured to: and training the object category identification model by a back propagation mode according to the text information of the sample service object and the category to which the sample service object belongs.
FIG. 14 illustrates an internal block diagram of a computer device in one embodiment. The computer device may in particular be the terminal in fig. 1. As shown in fig. 14, the computer device includes a processor, a memory, and a network interface connected by a system bus. The memory includes a nonvolatile storage medium and an internal memory. The non-volatile storage medium of the computer device stores an operating system, and may also store a computer program that, when executed by a processor, causes the processor to implement a business object recommendation method. The internal memory may also store a computer program that, when executed by the processor, causes the processor to perform a business object recommendation method.
It will be appreciated by those skilled in the art that the structure shown in fig. 14 is merely a block diagram of a portion of the structure associated with the present application and is not limiting of the computer device to which the present application applies, and that a particular computer device may include more or fewer components than shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, the service object recommending apparatus provided in the present application may be implemented as a computer program, which may be executed on a computer device as shown in fig. 14. The memory of the computer device may store various program modules that make up the recommendation device for the business object, such as the acquisition module 1202 and the determination module 1204 shown in FIG. 12. The computer program constituted by the respective program modules causes the processor to execute the steps in the business object recommendation method of the respective embodiments of the present application described in the present specification.
In one embodiment, a computer device is provided that includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the business object recommendation method described above. The step of the service object recommendation method herein may be a step in the service object recommendation method of each of the above embodiments.
In one embodiment, a storage medium is provided, storing a computer program which, when executed by a processor, causes the processor to perform the steps of the business object recommendation method described above. The step of the service object recommendation method herein may be a step in the service object recommendation method of each of the above embodiments.
Those skilled in the art will appreciate that all or part of the processes in the methods of the above embodiments may be implemented by a computer program for instructing relevant hardware, where the program may be stored in a non-volatile computer readable storage medium, and where the program, when executed, may include processes in the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the various embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples only represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the present application. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application is to be determined by the claims appended hereto.
Claims (13)
1. A method for recommending business objects, the method comprising:
acquiring text vectors of triggered target service objects, wherein the text vectors represent the relationship between the service objects, and the closer the relationship between the service objects is, the smaller the vector distance between the corresponding text vectors is;
acquiring text vectors of other business objects except the target business object, wherein the text vectors of the business objects are determined according to an object type recognition model obtained through training, the business objects comprise the target business object and the other business objects, the object type recognition model is obtained through training of text information of a sample business object and a type to which the sample business object belongs, the type to which the business object belongs is determined according to the text vectors of the business objects, the text information of the business object is preprocessed before the text information of the business object is input into the object type recognition model, and numerical representation of the text information of the business object is acquired, and the numerical representation of the text information of the business object is determined through a pre-constructed mapping table;
And determining a service object to be recommended according to the text vector of the target service object and the text vectors of the other service objects, and outputting the service object to be recommended.
2. The method of claim 1, wherein obtaining a text vector of a target business object corresponding to a user operation behavior comprises:
acquiring a pre-stored matching relation between a business object and a text vector, wherein the matching relation is established according to the business object and the text vector of the business object determined by the object type recognition model;
and searching the text vector of the target business object corresponding to the user operation behavior according to the matching relation.
3. The method of claim 1, wherein obtaining a text vector of a target business object corresponding to a user operation behavior comprises:
acquiring text information of a target business object corresponding to the user operation behavior;
and inputting the text information of the target business object into the object category identification model to obtain the text vector of the target business object.
4. A method according to claim 3, wherein the object class identification model comprises an implicit layer;
Inputting the text information of the target business object into the object category recognition model to obtain a text vector of the target business object, wherein the text vector comprises the following components:
and inputting the text information of the target business object into the object category identification model to obtain the text vector of the target business object output by a preset layer in the hidden layer.
5. The method of claim 4, wherein the hidden layer comprises a fully connected layer;
inputting the text information of the target service object into the object category recognition model to obtain a text vector of the target service object output by a preset layer in the hidden layer, wherein the text vector comprises the following components:
and inputting the text information of the target business object into the object category identification model to obtain the text vector of the target business object output by the last full-connection layer in the hidden layers.
6. The method of claim 1, wherein the determining the business object to be recommended based on the text vector of the target business object and the text vectors of the other business objects comprises:
obtaining the similarity between the text vector of the target service object and the text vectors of the other service objects;
And determining the business object to be recommended from the other business objects according to the similarity.
7. The method of claim 6, wherein the obtaining the similarity between the text vector of the target business object and the text vectors of the other business objects comprises:
and obtaining vector distances between the text vectors of the target business object and the text vectors of the other business objects, and utilizing the vector distances to represent the similarity between the text vectors of the target business object and the text vectors of the other business objects.
8. The method of claim 6, wherein said determining the business object to be recommended among the other business objects based on the similarity comprises:
sorting the other business objects by using the similarity;
and determining the business object to be recommended according to the sequencing result.
9. The method of claim 8, wherein the determining the business object to be recommended according to the ranking result comprises:
and selecting a preset number of other service objects according to the sequence from the high similarity to the low similarity according to the sequencing result, and taking the selected other service objects as the service objects to be recommended.
10. The method of claim 1, wherein the training mode of the object class identification model comprises:
acquiring text information of a sample service object and a category to which the sample service object belongs;
and training the object category identification model by a back propagation mode according to the text information of the sample service object and the category to which the sample service object belongs.
11. A recommendation device for a business object, the device comprising:
the acquisition module is used for acquiring text vectors of the triggered target business objects, wherein the text vectors represent the relationship between the business objects, and the closer the relationship between the business objects is, the smaller the vector distance between the corresponding text vectors is;
the obtaining module is further configured to obtain text vectors of other service objects except the target service object, where the text vectors of the service objects are determined according to an object class identification model obtained by training, the service objects include the target service object and the other service objects, the object class identification model is obtained by training text information of a sample service object and a class to which the sample service object belongs, determine a class to which the service object belongs according to the text vectors of the service object, and pre-process the text information of the service object before inputting the text information of the service object into the object class identification model, obtain a numerical representation of the text information of the service object, and the numerical representation of the text information of the service object is determined by a mapping table constructed in advance;
And the determining module is used for determining the service object to be recommended according to the text vector of the target service object and the text vectors of the other service objects and outputting the service object to be recommended.
12. A computer device comprising a memory and a processor, the memory having stored therein a computer program which, when executed by the processor, causes the processor to perform the steps of the method of any of claims 1 to 10.
13. A storage medium having stored thereon computer executable instructions which, when executed by a processor, cause the processor to perform the steps of the method of any of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910973183.8A CN110737834B (en) | 2019-10-14 | 2019-10-14 | Recommendation method and device for business objects, storage medium and computer equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910973183.8A CN110737834B (en) | 2019-10-14 | 2019-10-14 | Recommendation method and device for business objects, storage medium and computer equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110737834A CN110737834A (en) | 2020-01-31 |
| CN110737834B true CN110737834B (en) | 2024-04-02 |
Family
ID=69268851
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910973183.8A Active CN110737834B (en) | 2019-10-14 | 2019-10-14 | Recommendation method and device for business objects, storage medium and computer equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110737834B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11783707B2 (en) | 2018-10-09 | 2023-10-10 | Ford Global Technologies, Llc | Vehicle path planning |
| US11460851B2 (en) | 2019-05-24 | 2022-10-04 | Ford Global Technologies, Llc | Eccentricity image fusion |
| US11521494B2 (en) | 2019-06-11 | 2022-12-06 | Ford Global Technologies, Llc | Vehicle eccentricity mapping |
| US11662741B2 (en) | 2019-06-28 | 2023-05-30 | Ford Global Technologies, Llc | Vehicle visual odometry |
| CN113343711B (en) * | 2021-06-29 | 2024-05-10 | 南方电网数字电网研究院有限公司 | Work order generation method, device, equipment and storage medium |
| CN113505304B (en) * | 2021-09-10 | 2021-12-17 | 明品云(北京)数据科技有限公司 | Target object recommendation method and system |
| CN113850384A (en) * | 2021-09-30 | 2021-12-28 | 维沃移动通信有限公司 | Model training method and device |
| US12046047B2 (en) | 2021-12-07 | 2024-07-23 | Ford Global Technologies, Llc | Object detection |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106296195A (en) * | 2015-05-29 | 2017-01-04 | 阿里巴巴集团控股有限公司 | A kind of Risk Identification Method and device |
| WO2018099275A1 (en) * | 2016-11-29 | 2018-06-07 | 阿里巴巴集团控股有限公司 | Method, apparatus, and system for generating business object attribute identifier |
| WO2019105432A1 (en) * | 2017-11-29 | 2019-06-06 | 腾讯科技（深圳）有限公司 | Text recommendation method and apparatus, and electronic device |
| CN110020112A (en) * | 2017-09-25 | 2019-07-16 | 北京京东尚科信息技术有限公司 | Object Push method and its system |
| CN110209922A (en) * | 2018-06-12 | 2019-09-06 | 中国科学院自动化研究所 | Object recommendation method, apparatus, storage medium and computer equipment |
| CN110309427A (en) * | 2018-05-31 | 2019-10-08 | 腾讯科技（深圳）有限公司 | A kind of object recommendation method, apparatus and storage medium |
-
2019
- 2019-10-14 CN CN201910973183.8A patent/CN110737834B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106296195A (en) * | 2015-05-29 | 2017-01-04 | 阿里巴巴集团控股有限公司 | A kind of Risk Identification Method and device |
| WO2018099275A1 (en) * | 2016-11-29 | 2018-06-07 | 阿里巴巴集团控股有限公司 | Method, apparatus, and system for generating business object attribute identifier |
| CN110020112A (en) * | 2017-09-25 | 2019-07-16 | 北京京东尚科信息技术有限公司 | Object Push method and its system |
| WO2019105432A1 (en) * | 2017-11-29 | 2019-06-06 | 腾讯科技（深圳）有限公司 | Text recommendation method and apparatus, and electronic device |
| CN110309427A (en) * | 2018-05-31 | 2019-10-08 | 腾讯科技（深圳）有限公司 | A kind of object recommendation method, apparatus and storage medium |
| CN110209922A (en) * | 2018-06-12 | 2019-09-06 | 中国科学院自动化研究所 | Object recommendation method, apparatus, storage medium and computer equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110737834A (en) | 2020-01-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110737834B (en) | Recommendation method and device for business objects, storage medium and computer equipment | |
| CN111784455B (en) | Article recommendation method and recommendation equipment | |
| Chakraborty et al. | Fashion recommendation systems, models and methods: A review | |
| Zhu et al. | Heterogeneous hypergraph embedding for document recommendation | |
| Zhang et al. | Enabling kernel-based attribute-aware matrix factorization for rating prediction | |
| CN107357793B (en) | Information recommendation method and device | |
| KR101419504B1 (en) | System and method providing a suited shopping information by analyzing the propensity of an user | |
| CN110008397B (en) | Recommendation model training method and device | |
| Ullah et al. | Image-based service recommendation system: A JPEG-coefficient RFs approach | |
| CN107209762A (en) | Visual interactive formula is searched for | |
| CN112488781A (en) | Search recommendation method and device, electronic equipment and readable storage medium | |
| Hu et al. | Examining collaborative filtering algorithms for clothing recommendation in e-commerce | |
| CN111967924A (en) | Commodity recommendation method, commodity recommendation device, computer device, and medium | |
| Hiriyannaiah et al. | Deep visual ensemble similarity (DVESM) approach for visually aware recommendation and search in smart community | |
| Goti et al. | Artificial Intelligence in Business-to-Customer Fashion Retail: A Literature Review | |
| Zeng et al. | Collaborative filtering via heterogeneous neural networks | |
| Patel et al. | CNNRec: Convolutional Neural Network based recommender systems-A survey | |
| Ren et al. | A survey of recommendation techniques based on offline data processing | |
| Bhuvaneshwari et al. | Top-n recommendation system using explicit feedback and outer product based residual cnn | |
| Nazari et al. | Scalable and data-independent multi-agent recommender system using social networks analysis | |
| Zhao et al. | Personalized recommendation by exploring social users’ behaviors | |
| Nasir et al. | A survey and taxonomy of sequential recommender systems for e-commerce product recommendation | |
| Gahier et al. | Cross Domain Recommendation Systems using Deep Learning: A Systematic Literature Review | |
| Zhang et al. | Multi-view dynamic heterogeneous information network embedding | |
| KR20210041733A (en) | Method, apparatus and computer program for fashion item recommendation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 40021459 Country of ref document: HK |
|
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |