CN111914159A - Information recommendation method and terminal - Google Patents
Information recommendation method and terminal Download PDFInfo
- Publication number
- CN111914159A CN111914159A CN201910389789.7A CN201910389789A CN111914159A CN 111914159 A CN111914159 A CN 111914159A CN 201910389789 A CN201910389789 A CN 201910389789A CN 111914159 A CN111914159 A CN 111914159A
- Authority
- CN
- China
- Prior art keywords
- information
- user
- target
- behavior data
- scoring
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000012545 processing Methods 0.000 claims abstract description 42
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 40
- 238000001914 filtration Methods 0.000 claims abstract description 39
- 238000002372 labelling Methods 0.000 claims abstract description 32
- 230000006399 behavior Effects 0.000 claims description 122
- 239000013598 vector Substances 0.000 claims description 73
- 239000011159 matrix material Substances 0.000 claims description 34
- 238000004422 calculation algorithm Methods 0.000 claims description 16
- 238000004590 computer program Methods 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 13
- 238000010801 machine learning Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 4
- 238000011176 pooling Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 3
- 238000003062 neural network model Methods 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000013138 pruning Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 101100545272 Caenorhabditis elegans zif-1 gene Proteins 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
Abstract
The invention is suitable for the technical field of computers, and provides an information recommendation method and a terminal, wherein the method comprises the following steps: when information browsed by a user is acquired, labeling the information to acquire target information; extracting text features of the label type through a preset convolutional neural network model; grading the text characteristics and the target historical behavior data through a preset collaborative filtering model to obtain a grading result; and determining target information to be recommended according to the grading result, and recommending the target information to the user. According to the embodiment of the invention, the text characteristics and the historical behavior data of the user are processed through the preset convolutional neural network model and the preset collaborative filtering model, and the information is recommended to the user according to the processing result, so that the user can quickly acquire interesting, valuable and targeted information.
Description
Technical Field
The invention belongs to the technical field of computers, and particularly relates to an information recommendation method and a terminal.
Background
With the rapid development of the internet, in this big data era, users are exposed to massive information every day, and meanwhile, providing high-precision and personalized information recommendation is more and more important. The information is information that the user can bring value to himself in a relatively short time because he obtains it in time and uses it.
However, the conventional information recommendation system recommends and presents the same kind of information to all users based on the time line (i.e., the conventional information recommendation system updates the information according to the time period). Because the information updating period is long, and the recommended information is the same for all users, the users cannot quickly acquire interesting, valuable and targeted information.
Disclosure of Invention
In view of this, embodiments of the present invention provide an information recommendation method and a terminal, so as to solve the problem in the prior art that a conventional information recommendation system recommends and displays information of the same kind to all users based on a timeline, and since an information update cycle is long and the recommended information is the same for all users, the users cannot quickly obtain interesting, valuable and targeted information.
A first aspect of an embodiment of the present invention provides an information recommendation method, including:
when information browsed by a user is acquired, labeling the information to acquire target information; the target information comprises a label type used for identifying the category of the information;
extracting text features of the label type through a preset convolutional neural network model;
grading the text characteristics and the target historical behavior data through a preset collaborative filtering model to obtain a grading result; the target historical behavior data is historical behavior data associated with the text features in the historical behavior data of the user; the scoring result is used for expressing the interest degree of the user in the information corresponding to the tag type;
and determining target information to be recommended according to the grading result, and recommending the target information to the user.
A second aspect of an embodiment of the present invention provides a terminal, including:
the processing unit is used for labeling the information to obtain target information when the information browsed by the user is obtained; the target information comprises a label type used for identifying the category of the information;
the extraction unit is used for extracting the text features of the label type through a preset convolutional neural network model;
the scoring unit is used for scoring the text characteristics and the target historical behavior data through a preset collaborative filtering model to obtain a scoring result; the target historical behavior data is historical behavior data associated with the text features in the historical behavior data of the user; the scoring result is used for expressing the interest degree of the user in the information corresponding to the tag type;
and the recommending unit is used for determining target information to be recommended according to the grading result and recommending the target information to the user.
A third aspect of an embodiment of the present invention provides another terminal, including a processor, an input device, an output device, and a memory, where the processor, the input device, the output device, and the memory are connected to each other, where the memory is used to store a computer program that supports the terminal to execute the above method, where the computer program includes program instructions, and the processor is configured to call the program instructions and execute the following steps:
when information browsed by a user is acquired, labeling the information to acquire target information; the target information comprises a label type used for identifying the category of the information;
extracting text features of the label type through a preset convolutional neural network model;
grading the text characteristics and the target historical behavior data through a preset collaborative filtering model to obtain a grading result; the target historical behavior data is historical behavior data associated with the text features in the historical behavior data of the user; the scoring result is used for expressing the interest degree of the user in the information corresponding to the tag type;
and determining target information to be recommended according to the grading result, and recommending the target information to the user.
A fourth aspect of embodiments of the present invention provides a computer-readable storage medium storing a computer program which, when executed by a processor, implements the steps of:
when information browsed by a user is acquired, labeling the information to acquire target information; the target information comprises a label type used for identifying the category of the information;
extracting text features of the label type through a preset convolutional neural network model;
grading the text characteristics and the target historical behavior data through a preset collaborative filtering model to obtain a grading result; the target historical behavior data is historical behavior data associated with the text features in the historical behavior data of the user; the scoring result is used for expressing the interest degree of the user in the information corresponding to the tag type;
and determining target information to be recommended according to the grading result, and recommending the target information to the user.
According to the embodiment of the invention, when information browsed by a user is acquired, labeling is carried out on the information to obtain target information; the target information comprises a label type used for identifying the category of the information; extracting text features of the label type through a preset convolutional neural network model; grading the text characteristics and the target historical behavior data through a preset collaborative filtering model to obtain a grading result; the target historical behavior data is historical behavior data associated with the text features in the historical behavior data of the user; the scoring result is used for expressing the interest degree of the user in the information corresponding to the tag type; and determining target information to be recommended according to the grading result, and recommending the target information to the user. According to the embodiment of the invention, the text features of the label types are extracted through the preset convolutional neural network model, the text features and the historical behavior data of the user are processed through the preset collaborative filtering model, and the target recommendation information is determined according to the scoring result obtained through processing and is recommended to the user. Because the text characteristics and the historical behavior data of the user are processed through the preset neural network model and the preset collaborative filtering model, the data processing speed is improved, and the information recommendation rate is further improved; different information is recommended for each user, so that the user can quickly acquire interesting, valuable and targeted information.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the embodiments or the prior art descriptions will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive exercise.
Fig. 1 is a flowchart illustrating an implementation of an information recommendation method according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating an implementation of an information recommendation method according to another embodiment of the present invention;
fig. 3 is a schematic diagram of a terminal according to an embodiment of the present invention;
fig. 4 is a schematic diagram of a terminal according to another embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 1, fig. 1 is a schematic flow chart of an information recommendation method according to an embodiment of the present invention. The execution subject of the information recommendation method in this embodiment is a terminal, and the terminal includes but is not limited to a mobile terminal such as a smart phone, a tablet computer, a Personal Digital Assistant (PDA), and the like, and may also include a terminal such as a desktop computer. The information recommendation method as shown in fig. 1 may include:
s101: when information browsed by a user is acquired, labeling the information to acquire target information; the target information comprises a label type used for identifying the category of the information.
The information is information that the user can bring value to himself in a relatively short time because he obtains it in time and uses it. The information may be a text, an article, a sentence, etc., and the content, form, data amount, format, etc. of the information are not limited. The labeling processing refers to labeling the information; the label is used for identifying the category to which the information belongs.
When the terminal acquires the information browsed by the user, the information is labeled to obtain target information. Specifically, the information browsed by the user may be information uploaded to the terminal by the user, or may be information obtained by the terminal automatically when the terminal detects that the user browses the information. The information is labeled by a preset label labeling model; the label labeling model is obtained by training an information sample set by using a machine learning algorithm; labeling the information by a text classifier; the information can be labeled in a manual marking mode.
Labeling the information to obtain target information; the target information comprises a label type, and the label type is used for identifying the category of the information. The number of the tag types in one object information may be one or more, but is not limited thereto.
Further, S101 may specifically include: labeling the information through a preset label labeling model to obtain target information; the label labeling model is obtained by training an information sample set by using a machine learning algorithm.
The preset label labeling model is obtained by training an information sample set by using a machine learning algorithm. The information sample set may be a plurality of labeled information, and the number of labels may be one or more, which is not limited herein. It can be known that, the more information in the information sample set, the higher the accuracy of the trained model. Specifically, a text classifier is used for carrying out automatic label labeling training on the information sample set to generate a label labeling model.
And inputting the acquired information browsed by the user into a trained label labeling model, and labeling the information by the label labeling model to obtain labeled target information. The output target information may be:
information 1, tag 2, tag 3 … …
Information 2, tag 4, tag 5, tag 6 … …
The target information is only an exemplary illustration, the content of the tag in different target information may be the same or different, and the number of the tags in the target information may be one or more, which is not limited herein.
S102: and extracting the text features of the label type through a preset convolutional neural network model.
The preset convolutional neural network model is obtained by training a label type sample set by using a machine learning algorithm, in the training process, the input of the convolutional neural network model is information containing the label type, and the output of the convolutional neural network model is the extracted text feature of the label type. The tag type sample set may be composed of information including tag types, or may be directly composed of tag types.
And extracting the text features of the label type through a preset convolutional neural network model. Specifically, the convolutional neural network model converts the tag type into a vector matrix corresponding to the tag type; extracting a characteristic vector of the vector matrix through a convolutional neural network model; further, representative features, namely text features, are extracted from the feature vectors through a convolutional neural network model.
Further, in order to make the extracted text features more accurate and have more context features, and speed up the processing speed of extracting the text features, S102 may include: S1021-S1023 are as follows:
s1021: and inputting the label type into the convolutional neural network model for processing to obtain a vector matrix corresponding to the label type.
The convolutional neural network model may include an input layer (embedding layer), a convolutional layer, a pooling layer, and an output layer. The input layer comprises an input layer node, and the input layer node is used for receiving an input label type and converting the label type into a vector matrix corresponding to the label type. Each label type has a vector matrix corresponding to the label type after conversion; when the label types are multiple, converting to obtain multiple vector matrixes; the convolutional neural network model transfers the vector matrix obtained by conversion to the convolutional layer through the input layer.
S1022: and extracting the characteristic vector of the vector matrix.
The convolution layer in the convolutional neural network model processes the vector matrix and extracts the local features of the vector matrix, namely the feature vectors. When the number of the vector matrixes is one, the convolution layer convolves the whole area of the vector matrixes to obtain feature vectors; when the vector matrixes are multiple, according to the sequence of inputting the vector matrixes into the convolutional layer, firstly, the input first vector matrix is convolved to obtain a feature vector, then, a candidate area is generated on the feature vector, and the convolution is carried out on the candidate area and the next vector matrix to obtain a new feature vector. The convolution of the vector matrices continues in this manner until all the vector matrices are convolved, and the final output eigenvectors are passed to the pooling layer.
S1023: extracting the text features of the feature vector.
And processing the feature vectors by a pooling layer in the convolutional neural network model, and further extracting more representative features from the feature vectors to obtain text features. The pooling layer includes a maximum pooling layer node for extracting the maximum value in the feature vector. Specifically, the pooling layer performs normalization processing on the feature vectors, screens out the maximum value of the feature vectors, and obtains text features.
Further, in order to make the extracted text features more accurate and accelerate the processing speed of extracting the text features, the pooling layer may construct a feature vector of a fixed length through pooling operation, i.e., the feature vector obtained by the convolutional layer is processed into a feature vector of a fixed length. Specifically, the maximum feature value may be extracted from each feature vector by using a down-sampling (sampling point number reduction) method, the feature vector may be processed into a feature vector with a fixed length, and the text feature in the feature vector may be extracted.
S103: grading the text characteristics and the target historical behavior data through a preset collaborative filtering model to obtain a grading result; the target historical behavior data is historical behavior data associated with the text features in the historical behavior data of the user; and the scoring result is used for expressing the interest degree of the user in the information corresponding to the tag type.
The preset collaborative filtering model is obtained by training a training sample set by using a machine learning algorithm. Wherein the training sample set includes a plurality of sets of text features and historical behavior data associated with each set of text features. In the training process, the input of the collaborative filtering model is the text feature and the historical behavior data associated with the text feature, and the output of the collaborative filtering model is a scoring result obtained by scoring the text feature and the historical behavior data.
The historical behavior data refers to data generated by operating the information within preset time by a user; the preset time is set by the user according to actual conditions, for example, the preset time may be 48 hours, 24 hours, 12 hours, 1 hour, 30 minutes, and the like, which is not limited. The historical behavior data is offline data, and when a user operates the information, the generated data can generate a log file, and the log file within the preset time (namely, the offline data) is obtained and stored.
The specific operation behavior of the user on the information can be browsing, collecting, praise, marking dislike, no longer appearing, sharing, commenting, downloading, forwarding, searching, browsing times, browsing duration, adding a favorite list, removing the favorite list, switching the information and the like. Furthermore, in order to facilitate subsequent calculation of scores through the collaborative filtering model, corresponding weights can be set for each behavior in the user historical behavior data, and the weights and text features can be conveniently scored by the subsequent collaborative filtering model.
The target historical behavior data refers to historical behavior data associated with text features in the historical behavior data of the user. Specifically, each user has a user identification information, such as a user ID, and an information identifier is set for each information, the information identifier is used for identifying which information is specific, and the information identifier is associated with the user ID. The log file may include a user ID and an information identifier associated with the user ID, specific content of the information, historical behavior data of the user, and the like.
The text features obtained by processing the information browsed by the user can comprise information identification, a user ID associated with the information identification is searched in a log file according to the information identification, and historical behavior data generated by the operation of the user on the information is searched according to the user ID. For example, when a certain information is marked as A, the information is processed to obtain text characteristics, and target historical behavior data is searched in a log, namely data generated by the operation of a user on the information marked as A by the information is searched.
Furthermore, in order to improve the accuracy of recommending information and increase the processing speed, the historical behavior data of the user can be screened, and the screened historical behavior data is associated with the text features. For example, the user may perform similar operations on the information, and only one of them may be reserved.
And scoring the text characteristics and the target historical behavior data through a preset collaborative filtering model to obtain a scoring result. Specifically, converting the acquired target historical behavior data into a vector corresponding to the target historical behavior data; constructing a scoring matrix for the text features and the vectors through a collaborative filtering model; and determining a grading result according to the grading matrix. The scoring result is used for representing the interest degree of the user in the information corresponding to the tag type, and the scoring result may include a scoring value and a tag type associated with the scoring value, and may further include a user ID and the like.
Further, in order to accurately analyze the historical behavior data of the user and improve the accuracy of information recommendation, S103 may include: S1031-S1033, which is as follows:
s1031: and acquiring the target historical behavior data, and converting the target historical behavior data into a vector corresponding to the target historical behavior data.
The method comprises the steps that a terminal obtains target historical behavior data, specifically, text characteristics can include information identification, the terminal searches a user ID (identity) related to the information identification in a log file according to the information identification, and searches historical behavior data generated by operation of a user on the information according to the user ID; or the user uploads the target historical behavior data associated with the text features to the terminal, and the terminal searches the target historical behavior data associated with the text features according to the text features.
The collaborative filtering model may include an input layer, a processing layer, and an output layer. The input layer includes an input layer node for receiving input text features and target historical behavior data. After target historical behavior data are input into a collaborative filtering model, the collaborative filtering model converts the target historical behavior data into a vector corresponding to the target historical behavior data, and transmits the vector and the text features to a processing layer.
S1032: and constructing a scoring matrix based on the text features and the vectors.
The stochastic gradient descent algorithm is a model-free optimization algorithm, is suitable for training with more control variables and more complex controlled systems and cannot establish accurate mathematical models; hidden vectors refer to vectors of random variables that are not observable.
And constructing a scoring matrix based on the text features and the vectors, namely constructing a hidden vector matrix through a collaborative filtering model. Specifically, the hidden vector matrix is trained through a stochastic gradient descent algorithm, and can comprise a user hidden vector matrix and an information hidden vector matrix, wherein the user hidden vector matrix is obtained through calculation based on the stochastic gradient descent algorithm and text features, and the information hidden vector matrix is obtained through calculation based on the stochastic gradient descent algorithm and vectors corresponding to target historical behavior data.
S1033: and determining a scoring result based on the scoring matrix.
And performing point multiplication on data in the user hidden vector matrix and the information hidden vector matrix layer by layer in the collaborative filtering model to obtain a predicted grading result. The scoring result is used for representing the interest degree of the user in the information corresponding to the tag type in the scoring result, and the scoring result may include a scoring value and the associated tag type, and may further include a user ID and the like. The scoring result may include one or more scoring values and a tag type associated with each scoring value.
For example, the scoring result is user 1-stock securities-90; user 1-entertainment news-60; user 1-fund-80, etc.; wherein the values 90, 60, 80 are scoring values, which reflect the degree of interest of the user in the information corresponding to the tag type. The user can adjust and set the score rule according to the actual situation, such as the total score is 100, the total score is 10, the total score is 1000, and the like; the larger the score is, the higher the user interest degree is, the smaller the score is, the lower the user interest degree is, or the larger the score is, the lower the user interest degree is, and the smaller the score is, the higher the user interest degree is; the information corresponding to the label type in the top order is selected when the scores are the same, or the information corresponding to the label types with the same scores is recommended, and the like, which is not limited.
S104: and determining target information to be recommended according to the grading result, and recommending the target information to the user.
The target information is information which needs to be recommended to the user and is determined according to the grading result. Specifically, the scoring results may be ranked according to the score values from high to low, or from high to low, to obtain the scoring result with the highest score value, and obtain the tag type and the user ID in the scoring result, determine the target information to be recommended according to the tag type in the scoring result, and recommend the target information to the user according to the user ID.
Furthermore, the scoring result may further include information corresponding to the tag type in the scoring result. And when a scoring result with the maximum scoring value is obtained, extracting information in the scoring result as target information, and recommending the target information to the user according to the user ID in the scoring result.
Further, when the scoring result includes a scoring value and a tag type associated therewith, S104 may include: S1041-S1042, which is as follows:
s1041: and identifying the label type associated with the highest scoring value as a target label type based on the scoring value and the label type associated with the scoring value in the scoring result.
And sorting the scoring results based on the scoring values in the scoring results according to a sorting method preset by a user. The sorting method is preset by a user, and may be set to sort the scoring values from low to high, or may be set to sort the scoring values from high to low, and the like, which is not limited herein.
When the ranking method preset by the user is that the scoring values are ranked from high to low, the tag type associated with the first scoring value (namely the highest scoring value) is obtained, and the tag type is identified and marked as the target tag type.
S1042: and searching target information corresponding to the type of the target label, and recommending the target information to the user.
And searching target information corresponding to the target label type in a local database or a server according to the target label type. Specifically, a plurality of label types and information corresponding to each label type are stored in the local database and the server, and the terminal can search target information corresponding to the target note type in the local database and the server according to the target label type; or searching the target information corresponding to the target label type in the network according to the target label type, and acquiring the specific content of the target information and recommending the content to the user when the target information is searched.
Further, when the target information is searched, the identifier, link or address of the target information may be obtained, and the information such as the identifier, link or address may be recommended to the user. The user can check the information according to the information identifier, or check the information by clicking the link, or search the target information according to the address of the target information.
According to the embodiment of the invention, when information browsed by a user is acquired, labeling is carried out on the information to obtain target information; the target information comprises a label type used for identifying the category of the information; extracting text features of the label type through a preset convolutional neural network model; grading the text characteristics and the target historical behavior data through a preset collaborative filtering model to obtain a grading result; the target historical behavior data is historical behavior data associated with the text features in the historical behavior data of the user; the scoring result is used for expressing the interest degree of the user in the information corresponding to the tag type; and determining target information to be recommended according to the grading result, and recommending the target information to the user. According to the embodiment of the invention, the text features of the label types are extracted through the preset convolutional neural network model, the text features and the historical behavior data of the user are processed through the preset collaborative filtering model, and the target recommendation information is determined according to the scoring result obtained through processing and is recommended to the user. Because the text characteristics and the historical behavior data of the user are processed through the preset neural network model and the preset collaborative filtering model, the data processing speed is improved, and the information recommendation rate is further improved; different information is recommended for each user, so that the user can quickly acquire interesting, valuable and targeted information.
Referring to fig. 2, fig. 2 is a schematic flow chart of an information recommendation method according to another embodiment of the present invention. The execution subject of the information recommendation method in this embodiment is a terminal, and the terminal includes but is not limited to a mobile terminal such as a smart phone, a tablet computer, and a personal digital assistant, and may also include a terminal such as a desktop computer.
The difference between the present embodiment and the embodiment corresponding to fig. 1 is that S205-S206 may be further included after S204, where S201-S204 in the present embodiment are completely the same as S101-S104 in the previous embodiment, and reference is specifically made to the description related to S101-S104 in the previous embodiment, which is not repeated herein.
Further, in order to recommend more accurate information to the user in real time, S204 may be followed by S205-S206, which is as follows:
s205: and acquiring real-time behavior data of the user associated with the text features.
The real-time behavior data is different from the historical behavior data, the historical behavior data is offline data, the real-time behavior data is behavior data fed back by a user in real time, and it can be understood that when the user performs any behavior operation on information, the generated data is fed back to the terminal, and the data received by the terminal is the real-time behavior data. The text characteristics can comprise information marks, information operated by a user also comprises information marks, the text characteristics and the information are related according to the information marks, and when the user operates the information, real-time behavior data generated by user operation is related to the text characteristics.
The terminal can acquire the real-time behavior data of the user associated with the text features within a preset time. The general preset time setting for the real-time behavior data is short, such as 2 minutes, 3 minutes, 5 minutes and the like, and the user can adjust the real-time behavior data according to the actual situation without limitation.
S206: and inputting the scoring result and the real-time behavior data into the convolutional neural network model for data iteration processing, and updating the scoring result according to the data iteration processing result.
The convolutional neural network model may include a real-time stream recommendation module, and the real-time stream recommendation module is configured to process the scoring result and the real-time behavior data output by the collaborative filtering model, update the scoring result according to the processing result, and recommend target information based on the updated scoring result.
Specifically, the scoring result output by the collaborative filtering model and the acquired real-time behavior data are input into a real-time flow recommendation module in the convolutional neural network model, and the real-time flow recommendation module performs data iteration processing on the scoring result and the real-time behavior data to obtain a data iteration processing result. Iteration refers to the activity of repeating the feedback process, each repetition of the process becomes one iteration, and the result of each iteration is used as the initial value of the next iteration. The scoring result is updated in real time through data iteration processing, and all behavior data of the user do not need to be calculated and analyzed every time, so that the workload can be greatly reduced, and the information recommendation efficiency and accuracy are improved.
And after the grading result and the real-time behavior data are subjected to data iteration processing through the real-time stream recommendation module to obtain a data iteration processing result, updating the grading result according to the processing result. And searching corresponding target information according to the updated grading result, and recommending the target information to the user.
Further, in order to ensure real-time performance of data and improve accuracy and efficiency of information recommendation, when historical behavior data is processed through the collaborative filtering model and real-time behavior data is processed through the real-time stream recommendation module, a part of historical data, such as user behavior data with too long discarding time, can be discarded by using a sliding window, and latest real-time data information is reserved for calculation. The sliding window is a flow control technique for controlling data flow.
Further, in order to improve the rate and accuracy of information recommendation, the user, the information, the historical behavior data, and the like may be classified by using a clustering algorithm, for example, the information browsed by the user is classified according to the age, sex, occupation, liveness, academic calendar, hobby, income, and the like of the user, and the terminal processes the same type of data each time. Because the same type of data is processed every time, the accuracy of the convolutional neural network model and the collaborative filtering model for processing the data is higher and higher, and therefore information recommendation is more and more accurate. The clustering algorithm is a segmentation clustering method, and is used for searching representative data in a large amount of data.
Generally, it is considered that the behavior operation performed by the user on the information within a certain period of time, the generated behavior data is related to the information, and the period of time can be set by the user, and the time is generally short. Data generated by any behavior operation performed on information by a user needs to be processed by a convolutional neural network model and a collaborative filtering model, if the data is long, the significance of recommending the information is not great, and a terminal needs to spend a large amount of time on processing. At this time, when historical behavior data and real-time behavior data are processed through a convolutional neural network model and a collaborative filtering model, a part of historical behavior data can be discarded by using a pruning algorithm, and the information recommendation rate and accuracy are improved. The pruning algorithm avoids unnecessary traversal processes through a self-defined judgment rule.
According to the embodiment of the invention, when information browsed by a user is acquired, labeling is carried out on the information to obtain target information; the target information comprises a label type used for identifying the category of the information; extracting text features of the label type through a preset convolutional neural network model; grading the text characteristics and the target historical behavior data through a preset collaborative filtering model to obtain a grading result; the target historical behavior data is historical behavior data associated with the text features in the historical behavior data of the user; the scoring result is used for expressing the interest degree of the user in the information corresponding to the tag type; and determining target information to be recommended according to the grading result, and recommending the target information to the user. According to the embodiment of the invention, the text features of the label types are extracted through the preset convolutional neural network model, the text features and the historical behavior data of the user are processed through the preset collaborative filtering model, and the target recommendation information is determined according to the scoring result obtained through processing and is recommended to the user. Because the text characteristics and the historical behavior data of the user are processed through the preset neural network model and the preset collaborative filtering model, the data processing speed is improved, and the information recommendation rate is further improved; different information is recommended for each user, so that the user can quickly acquire interesting, valuable and targeted information.
Referring to fig. 3, fig. 3 is a schematic diagram of a terminal according to an embodiment of the present invention. The terminal includes units for executing the steps in the embodiments corresponding to fig. 1 and fig. 2. Please refer to fig. 1 and fig. 2 for the corresponding embodiments. For convenience of explanation, only the portions related to the present embodiment are shown. Referring to fig. 3, the terminal 3 includes:
the processing unit 310 is configured to, when information browsed by a user is acquired, perform tagging processing on the information to acquire target information; the target information comprises a label type used for identifying the category of the information;
an extracting unit 320, configured to extract a text feature of the tag type through a preset convolutional neural network model;
the scoring unit 330 is configured to score the text features and the target historical behavior data through a preset collaborative filtering model to obtain a scoring result; the target historical behavior data is historical behavior data associated with the text features in the historical behavior data of the user; the scoring result is used for expressing the interest degree of the user in the information corresponding to the tag type;
and the recommending unit 340 is configured to determine target information to be recommended according to the scoring result, and recommend the target information to the user.
Further, the extracting unit 320 is specifically configured to:
inputting the label type into the convolutional neural network model for processing to obtain a vector matrix corresponding to the label type;
extracting a characteristic vector of the vector matrix;
extracting the text features of the feature vector.
Further, the scoring unit 330 is specifically configured to:
acquiring the target historical behavior data, and converting the target historical behavior data into a vector corresponding to the target historical behavior data;
constructing a scoring matrix based on the text features and the vectors;
and determining a scoring result based on the scoring matrix.
Further, the scoring result includes a scoring value and a tag type associated therewith, and the recommending unit 340 is specifically configured to:
identifying a tag type associated with the highest scoring value as a target tag type based on the scoring value and its associated tag type in the scoring result;
and searching target information corresponding to the type of the target label, and recommending the target information to the user.
Further, the terminal further includes:
the acquiring unit is used for acquiring real-time behavior data of the user associated with the text features;
and the updating unit is used for inputting the scoring result and the real-time behavior data into the convolutional neural network model for data iteration processing, and updating the scoring result according to the data iteration processing result.
Further, the processing unit 310 is specifically configured to:
labeling the information through a preset label labeling model to obtain target information; the label labeling model is obtained by training an information sample set by using a machine learning algorithm.
Referring to fig. 4, fig. 4 is a schematic diagram of a terminal according to another embodiment of the present invention. As shown in fig. 4, the terminal 4 of this embodiment includes: a processor 40, a memory 41 and a computer program 42 stored in said memory 41 and executable on said processor 40. The processor 40 executes the computer program 42 to implement the steps in the above-described information recommendation method embodiments of each terminal, for example, S101 to S104 shown in fig. 1. Alternatively, the processor 40, when executing the computer program 42, implements the functions of the units in the device embodiments, such as the functions of the units 310 to 340 shown in fig. 4.
Illustratively, the computer program 42 may be divided into one or more units, which are stored in the memory 41 and executed by the processor 40 to accomplish the present invention. The one or more units may be a series of computer program instruction segments capable of performing specific functions, which are used to describe the execution of the computer program 42 in the terminal 4. For example, the computer program 42 may be divided into a processing unit, an extraction unit, a scoring unit, and a recommendation unit, each unit functioning as described above.
The terminal may include, but is not limited to, a processor 40, a memory 41. Those skilled in the art will appreciate that fig. 4 is merely an example of a terminal 4 and is not intended to be limiting of terminal 4, and may include more or fewer components than shown, or some components in combination, or different components, e.g., the terminal may also include input and output terminals, network access terminals, buses, etc.
The Processor 40 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic, discrete hardware components, etc. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory 41 may be an internal storage unit of the terminal 4, such as a hard disk or a memory of the terminal 4. The memory 41 may also be an external storage terminal of the terminal 4, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card) and the like provided on the terminal 4. Further, the memory 41 may also include both an internal storage unit of the terminal 4 and an external storage terminal. The memory 41 is used for storing the computer program and other programs and data required by the terminal. The memory 41 may also be used to temporarily store data that has been output or is to be output.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present invention, and are intended to be included within the scope of the present invention.
Claims (10)
1. An information recommendation method, comprising:
when information browsed by a user is acquired, labeling the information to acquire target information; the target information comprises a label type used for identifying the category of the information;
extracting text features of the label type through a preset convolutional neural network model;
grading the text characteristics and the target historical behavior data through a preset collaborative filtering model to obtain a grading result; the target historical behavior data is historical behavior data associated with the text features in the historical behavior data of the user; the scoring result is used for expressing the interest degree of the user in the information corresponding to the tag type;
and determining target information to be recommended according to the grading result, and recommending the target information to the user.
2. The information recommendation method of claim 1, wherein the extracting the text feature of the tag type through a preset convolutional neural network model comprises:
inputting the label type into the convolutional neural network model for processing to obtain a vector matrix corresponding to the label type;
extracting a characteristic vector of the vector matrix;
extracting the text features of the feature vector.
3. The information recommendation method according to claim 1, wherein the scoring the text features and the target historical behavior data through a preset collaborative filtering model to obtain a scoring result comprises:
acquiring the target historical behavior data, and converting the target historical behavior data into a vector corresponding to the target historical behavior data;
constructing a scoring matrix based on the text features and the vectors;
and determining a scoring result based on the scoring matrix.
4. The information recommendation method of claim 1, wherein the scoring result includes a scoring value and a tag type associated therewith, and the determining target information to be recommended according to the scoring result and recommending the target information to the user includes:
identifying a tag type associated with the highest scoring value as a target tag type based on the scoring value and its associated tag type in the scoring result;
and searching target information corresponding to the type of the target label, and recommending the target information to the user.
5. The information recommendation method according to any one of claims 1 to 4, wherein after scoring the text features and the target historical behavior data through a preset collaborative filtering model and obtaining a scoring result, the method further comprises:
acquiring real-time behavior data of the user associated with the text features;
and inputting the scoring result and the real-time behavior data into the convolutional neural network model for data iteration processing, and updating the scoring result according to the data iteration processing result.
6. The information recommendation method according to claim 1, wherein when the information browsed by the user is obtained, the tagging is performed on the information to obtain the target information, and the method comprises:
labeling the information through a preset label labeling model to obtain target information; the label labeling model is obtained by training an information sample set by using a machine learning algorithm.
7. A terminal, comprising:
the processing unit is used for labeling the information to obtain target information when the information browsed by the user is obtained; the target information comprises a label type used for identifying the category of the information;
the extraction unit is used for extracting the text features of the label type through a preset convolutional neural network model;
the scoring unit is used for scoring the text characteristics and the target historical behavior data through a preset collaborative filtering model to obtain a scoring result; the target historical behavior data is historical behavior data associated with the text features in the historical behavior data of the user; the scoring result is used for expressing the interest degree of the user in the information corresponding to the tag type;
and the recommending unit is used for determining target information to be recommended according to the grading result and recommending the target information to the user.
8. The terminal of claim 7, wherein the extraction unit is specifically configured to:
inputting the label type into the convolutional neural network model for processing to obtain a vector matrix corresponding to the label type;
extracting a characteristic vector of the vector matrix;
extracting the text features of the feature vector.
9. A terminal comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the steps of the method according to any of claims 1 to 6 when executing the computer program.
10. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910389789.7A CN111914159B (en) | 2019-05-10 | 2019-05-10 | Information recommendation method and terminal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910389789.7A CN111914159B (en) | 2019-05-10 | 2019-05-10 | Information recommendation method and terminal |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111914159A true CN111914159A (en) | 2020-11-10 |
| CN111914159B CN111914159B (en) | 2024-03-12 |
Family
ID=73242204
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910389789.7A Active CN111914159B (en) | 2019-05-10 | 2019-05-10 | Information recommendation method and terminal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111914159B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112348587A (en) * | 2020-11-16 | 2021-02-09 | 脸萌有限公司 | Information pushing method and device and electronic equipment |
| CN112417294A (en) * | 2020-12-03 | 2021-02-26 | 广州瀚信通信科技股份有限公司 | Intelligent business recommendation method based on neural network mining model |
| CN112948677A (en) * | 2021-02-26 | 2021-06-11 | 上海携旅信息技术有限公司 | Recommendation reason determination method, system, device and medium based on comment aesthetic feeling |
| WO2022116536A1 (en) * | 2020-12-04 | 2022-06-09 | 上海帜讯信息技术股份有限公司 | Information service providing method and apparatus, electronic device, and storage medium |
| CN116028617A (en) * | 2022-12-06 | 2023-04-28 | 腾讯科技（深圳）有限公司 | Information recommendation method, apparatus, device, readable storage medium and program product |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102609523A (en) * | 2012-02-10 | 2012-07-25 | 上海视畅信息科技有限公司 | Collaborative filtering recommendation algorithm based on article sorting and user sorting |
| US20150254371A1 (en) * | 2012-11-26 | 2015-09-10 | Linkedln Corporation | Techniques for inferring an organizational hierarchy from a social graph |
| CN107436950A (en) * | 2017-08-07 | 2017-12-05 | 苏州大学 | A kind of itinerary recommends method and system |
| CN107562875A (en) * | 2017-08-31 | 2018-01-09 | 北京麒麟合盛网络技术有限公司 | A kind of update method of model, apparatus and system |
| CN107665254A (en) * | 2017-09-30 | 2018-02-06 | 济南浪潮高新科技投资发展有限公司 | A kind of menu based on deep learning recommends method |
| CN108090229A (en) * | 2018-01-10 | 2018-05-29 | 广东工业大学 | A kind of method and apparatus that rating matrix is determined based on convolutional neural networks |
| CN108804577A (en) * | 2018-05-24 | 2018-11-13 | 联通在线信息科技有限公司 | A kind of predictor method of information label interest-degree |
| US20180365577A1 (en) * | 2016-05-12 | 2018-12-20 | Tencent Technology (Shenzhen) Company Limited | Data recommendation method and device, and storage medium |
| CN109388731A (en) * | 2018-08-31 | 2019-02-26 | 昆明理工大学 | A kind of music recommended method based on deep neural network |
-
2019
- 2019-05-10 CN CN201910389789.7A patent/CN111914159B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102609523A (en) * | 2012-02-10 | 2012-07-25 | 上海视畅信息科技有限公司 | Collaborative filtering recommendation algorithm based on article sorting and user sorting |
| US20150254371A1 (en) * | 2012-11-26 | 2015-09-10 | Linkedln Corporation | Techniques for inferring an organizational hierarchy from a social graph |
| US20180365577A1 (en) * | 2016-05-12 | 2018-12-20 | Tencent Technology (Shenzhen) Company Limited | Data recommendation method and device, and storage medium |
| CN107436950A (en) * | 2017-08-07 | 2017-12-05 | 苏州大学 | A kind of itinerary recommends method and system |
| CN107562875A (en) * | 2017-08-31 | 2018-01-09 | 北京麒麟合盛网络技术有限公司 | A kind of update method of model, apparatus and system |
| CN107665254A (en) * | 2017-09-30 | 2018-02-06 | 济南浪潮高新科技投资发展有限公司 | A kind of menu based on deep learning recommends method |
| CN108090229A (en) * | 2018-01-10 | 2018-05-29 | 广东工业大学 | A kind of method and apparatus that rating matrix is determined based on convolutional neural networks |
| CN108804577A (en) * | 2018-05-24 | 2018-11-13 | 联通在线信息科技有限公司 | A kind of predictor method of information label interest-degree |
| CN109388731A (en) * | 2018-08-31 | 2019-02-26 | 昆明理工大学 | A kind of music recommended method based on deep neural network |
Non-Patent Citations (2)
| Title |
|---|
| YUE SHI 等: "Exploiting user similarity based on rated-item pools for improved user-based collaborative filtering", 《RECSYS \'09: PROCEEDINGS OF THE THIRD ACM CONFERENCE ON RECOMMENDER SYSTEMS》, 23 October 2009 (2009-10-23), pages 125, XP058401305, DOI: 10.1145/1639714.1639736 * |
| 钟良月: "基于改进的个性化混合推荐算法的研究", 《中国优秀硕士学位论文全文数据库 社会科学Ⅱ辑》, no. 05, 15 May 2018 (2018-05-15), pages 123 - 28 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112348587A (en) * | 2020-11-16 | 2021-02-09 | 脸萌有限公司 | Information pushing method and device and electronic equipment |
| CN112417294A (en) * | 2020-12-03 | 2021-02-26 | 广州瀚信通信科技股份有限公司 | Intelligent business recommendation method based on neural network mining model |
| WO2022116536A1 (en) * | 2020-12-04 | 2022-06-09 | 上海帜讯信息技术股份有限公司 | Information service providing method and apparatus, electronic device, and storage medium |
| CN112948677A (en) * | 2021-02-26 | 2021-06-11 | 上海携旅信息技术有限公司 | Recommendation reason determination method, system, device and medium based on comment aesthetic feeling |
| CN112948677B (en) * | 2021-02-26 | 2023-11-03 | 上海携旅信息技术有限公司 | Recommendation reason determining method, system, equipment and medium based on comment aesthetic feeling |
| CN116028617A (en) * | 2022-12-06 | 2023-04-28 | 腾讯科技（深圳）有限公司 | Information recommendation method, apparatus, device, readable storage medium and program product |
| CN116028617B (en) * | 2022-12-06 | 2024-02-27 | 腾讯科技（深圳）有限公司 | Information recommendation method, apparatus, device, readable storage medium and program product |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111914159B (en) | 2024-03-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112632385B (en) | Course recommendation method, course recommendation device, computer equipment and medium | |
| CN111914159B (en) | Information recommendation method and terminal | |
| CN108804512B (en) | Text classification model generation device and method and computer readable storage medium | |
| CN110147551B (en) | Multi-category entity recognition model training, entity recognition method, server and terminal | |
| CN111966914B (en) | Content recommendation method and device based on artificial intelligence and computer equipment | |
| CN108846097B (en) | User interest tag representation method, article recommendation device and equipment | |
| CN112307762B (en) | Search result sorting method and device, storage medium and electronic device | |
| CN110968684A (en) | Information processing method, device, equipment and storage medium | |
| CN109598307B (en) | Data screening method and device, server and storage medium | |
| CN111898739B (en) | Data screening model construction method, data screening method, device, computer equipment and storage medium based on meta learning | |
| CN111881671B (en) | Attribute word extraction method | |
| CN111160191A (en) | Video key frame extraction method and device and storage medium | |
| WO2022134586A1 (en) | Meta-learning-based target classification method and apparatus, device and storage medium | |
| CN110598869B (en) | Classification method and device based on sequence model and electronic equipment | |
| CN111984792A (en) | Website classification method and device, computer equipment and storage medium | |
| CN113722438A (en) | Sentence vector generation method and device based on sentence vector model and computer equipment | |
| CN108959550B (en) | User focus mining method, device, equipment and computer readable medium | |
| CN112699945A (en) | Data labeling method and device, storage medium and electronic device | |
| CN110968664A (en) | Document retrieval method, device, equipment and medium | |
| CN113704620B (en) | User tag updating method, device, equipment and medium based on artificial intelligence | |
| CN105164672A (en) | Content classification | |
| CN113342799B (en) | Data correction method and system | |
| CN111126626A (en) | Training method, device, server, platform and storage medium | |
| CN107368464B (en) | Method and device for acquiring bidding product information | |
| CN113139368B (en) | Text editing method and system |
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 |