CN113127604B - Comment text-based fine-grained item recommendation method and system - Google Patents
Comment text-based fine-grained item recommendation method and system Download PDFInfo
- Publication number
- CN113127604B CN113127604B CN202110505981.5A CN202110505981A CN113127604B CN 113127604 B CN113127604 B CN 113127604B CN 202110505981 A CN202110505981 A CN 202110505981A CN 113127604 B CN113127604 B CN 113127604B
- Authority
- CN
- China
- Prior art keywords
- user
- article
- comment text
- comment
- granularity
- 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 29
- 230000003993 interaction Effects 0.000 claims abstract description 39
- 239000011159 matrix material Substances 0.000 claims abstract description 34
- 235000019580 granularity Nutrition 0.000 claims abstract description 32
- 230000002452 interceptive effect Effects 0.000 claims abstract description 25
- 238000013528 artificial neural network Methods 0.000 claims abstract description 15
- 239000013598 vector Substances 0.000 claims description 20
- 238000012552 review Methods 0.000 claims description 17
- 238000013527 convolutional neural network Methods 0.000 claims description 11
- 230000010339 dilation Effects 0.000 claims description 11
- 238000000605 extraction Methods 0.000 claims description 7
- 238000000354 decomposition reaction Methods 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 4
- 238000004590 computer program Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 claims description 4
- 230000008034 disappearance Effects 0.000 claims description 2
- 238000004880 explosion Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 4
- 238000011176 pooling Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000003306 harvesting Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000012795 verification 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
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Databases & Information Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The disclosure provides a comment text-based fine-grained item recommendation method and a comment text-based fine-grained item recommendation system, which comprise the following steps: respectively acquiring a user comment text set and an article comment text set; calculating a multi-granularity incidence matrix between a user and an article by using a fine-granularity feature interaction network to obtain a 3D interaction image; inputting the 3D interactive image into a fully-connected neural network and a traditional factorization machine to realize the grading prediction of a user-article, and realizing the recommendation of the article according to the grading result; according to the scheme, the user and article comments are coded by a multi-level expansion convolution structure, loss of fine-grained information in the comments is avoided, feature interaction of the user and the article comments is established under multiple granularities, multi-grained information is fused and processed by using 3D convolution, relevant information under multiple granularities in the comments is effectively highlighted, and the reasonability and the accuracy of article recommendation are effectively improved.
Description
Technical Field
The disclosure belongs to the technical field of article recommendation, and particularly relates to a comment text-based fine-grained article recommendation method and system.
Background
The statements in this section merely provide background information related to the present disclosure and may not constitute prior art.
In many works, it has proven effective to use comment text to boost the effectiveness of recommendation systems. The comment-based recommendation system not only alleviates the cold start problem, but also allows for finer grained user and item representations. Earlier work has focused primarily on topic modeling and language modeling approaches. In recent years, researchers have achieved good results using deep learning methods. ConvMF uses a Convolutional Neural Network (CNN) as an automatic feature extractor to encode users (items) into a low-dimensional vector representation. And then combined by Probability Matrix Factorization (PMF) to make score prediction. DeepCoNN (Deep collaborative Neural Networks) uses two parallel convolution architectures to infer potential feature representations from user and item reviews. The user and item feature representations are concatenated and used as input to a Factoring Machine (FM) for scoring predictions. Since different words have different importance to modeling, D-Attn introduces a word-level attention mechanism, giving different words different weights. DAML focuses on local information between users and item reviews and on interactions between the two. The DAML integrates the scoring feature and the comment feature into a unified frame, and realizes high-order nonlinear interaction of the features by using a nerve factor decomposition machine to complete final scoring prediction. The model interpretability is enhanced. The MPCN extracts important comments from the comments of users and items and then matches them word by word. The method can not only utilize the comments with the most abundant information to predict, but also carry out deeper character interaction.
With the great success of deep learning in the fields of natural language processing and the like, the performance of score prediction is greatly improved by a plurality of recommendation models based on the technology. In these works, the convolutional neural network architecture extracts potential features from corresponding reviews of users and items, respectively. The comments are represented by using a dense word embedding matrix, followed by application of a fixed-size sliding window to capture the contextual information. The convolutional neural network-based technology better promotes the understanding of semantic information in the comments, and the method leads to the remarkable improvement of the existing bag-of-words mode-based scoring prediction method. However, one of the significant disadvantages is that it is difficult for a person to effectively understand the features extracted by the neural network, which limits the interpretability of the recommendation system.
The key to a comment-based recommendation system is the capture of different granularity of information in the text of the comment. Generally speaking, the same user tends to have different preferences for different items, which collectively reflect the comments he writes. Meanwhile, important semantic features in the comments are hidden in text segments of different granularities. The illustration in the figure illustrates this problem. Different granularity of text (e.g., words, phrases, and sentences) may reveal user preferences for different items. For example, the word: "best", "great", "disapporting"; the phrase: "quality great battle life", "screen looks hooks amplitude" and the sentence: "I use all go with product read but it clicked a bit orange this year".
However, the inventor finds that the existing work usually learns a single feature vector for each user or article by integrating the scoring papers of the users and articles, and then realizes scoring prediction by means of factorization and the like. These methods have limitations in capturing fine-grained user preferences and item attributes because they simply assign a single feature vector to the user's item, resulting in information at multiple granularities in the review not being captured significantly.
Disclosure of Invention
In order to solve the problems, the invention provides a fine-grained article recommendation method and a fine-grained article recommendation system based on comment texts, the scheme provides a fine-grained feature interaction network, user and article comments are coded by a multi-level expansion convolution structure, loss of fine-grained information in the comments is avoided, feature interaction of the user and the article comments is constructed under multiple granularities, multi-grained information is fused and processed by using 3D convolution, relevant information under multiple granularities in the comments is effectively highlighted, and reasonability and accuracy of article recommendation are effectively improved.
According to a first aspect of the embodiments of the present disclosure, there is provided a comment text-based fine-grained item recommendation method, including:
respectively acquiring a user comment text set and an article comment text set;
calculating a multi-granularity incidence matrix between a user and an article by using a fine-granularity feature interaction network to obtain a 3D interaction image;
inputting the 3D interactive image into a fully-connected neural network and a traditional factorization machine to realize the grading prediction of a user-article, and realizing the recommendation of the article according to the grading result;
the fine-grained feature interaction network specifically includes: performing multi-granularity semantic feature extraction on each comment text in the user comment text set and the article comment text set by using a hierarchical expansion convolution model to obtain a multi-granularity comment expression set of the user and the article; and calculating a multi-granularity incidence matrix between the user and the article, and fusing the incidence matrix into a 3D interactive image through a 3D convolutional neural network.
Further, the convolution is performed on a continuous subsequence of the input at each step by using a hierarchical dilation convolution model different from the standard convolution, the dilation convolution at different levels is performed by skipping δ input elements at a time, and the dilation operation is specifically described as:
wherein,is vector concatenation, b is a bias term, reLU is a non-linear activation function, x t Representing the context center word, W is a convolution kernel of size 2w +1, W is a parameter for adjusting the size of the convolution kernel, and δ is an expansion rate, which is an integer not less than 1.
Further, the calculating a multi-granularity association matrix between the user and the article specifically includes: the interaction of features between users and articles is realized by constructing a correlation matrix on each granularity, and given the nth and mth multi-granularity comment representations of the users and the articles, the correlation matrix of the ith layer of granularity is realized by the following dot product operation:
wherein L = [1,2]As to the number of semantic granularities, element (1) ofRepresenting the relevance of the ith and jth features in the user and item reviews at the granularity of l.
According to a second aspect of the embodiments of the present disclosure, there is provided a comment text-based fine-grained item recommendation system, including:
the data acquisition unit is used for respectively acquiring a user comment text set and an article comment text set;
the multi-granularity feature extraction unit is used for respectively extracting multi-granularity semantic features of each comment text in the user comment text set and the article comment text set by utilizing the hierarchical expansion convolution model to obtain multi-granularity comment expression sets of the user and the article;
the interactive image generation unit is used for constructing a multi-granularity incidence matrix between the user and the article based on the multi-granularity comment expression of the user and the article, and fusing the incidence matrix into a 3D interactive image through a 3D convolutional neural network;
and the item recommending unit is used for inputting the 3D interactive image into the fully-connected neural network and the traditional factorization machine to realize the grading prediction of the user-item and realize the recommendation of the item according to the grading result.
According to a third aspect of the embodiments of the present disclosure, there is provided an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the memory, where the processor implements the comment text-based fine-grained item recommendation system when executing the program.
According to a fourth aspect of the embodiments of the present disclosure, there is provided a non-transitory computer-readable storage medium having stored thereon a computer program, which when executed by a processor, implements a comment text based fine-grained item recommendation system as described above.
Compared with the prior art, the beneficial effect of this disclosure is:
(1) The scheme of the present disclosure provides a fine-grained feature interaction network, and a multi-grained association matrix between a user and an article is calculated through the fine-grained feature interaction network, which has the following advantages: multi-level user/item representation and fine-grained feature interaction, the network does not represent each user by a single abstract vector, but uses hierarchical extended convolution to construct a comment-based multi-level representation in a unified module; by convolution with layered stack expansion, the received input width of each layer grows exponentially, while the number of parameters increases only linearly. At the same time, the output of each layer is retained as a feature map across text segments of different lengths, without loss of coverage because no form of pooling or stepping convolution is used; in this way, the scheme can step through local relevance and long-term dependency to obtain semantic features in comments, including word, phrase, and sentence hierarchies, at different granularities. According to the scheme, the user and item comments are coded by a multi-level expansion convolution structure, so that the loss of fine-grained information in the comments is avoided.
(2) In order to avoid information loss, the fine-grained feature interaction network constructs feature interaction of user comments and article comments on each semantic granularity. In practical application, the model is based on a hierarchical representation method of comment texts, and a correlation matrix is constructed for each pair of comments from word to sentence levels. In this way, a number of granular features implicit in user and item reviews may be identified and interacted with minimal loss, thereby providing sufficient content relevance cues for predicting accurate scores. Then, we merge multiple incidence matrices of each granular review pair into one 3D image, whose channels represent the degree of relevance of user and item review feature interactions at different granularities. Predicting a user's rating of the item by identifying a higher order salient signal similar to the 3D convolution based image identification hierarchy; according to the scheme, the characteristic interaction of the user and the item comment is constructed under multiple granularities, and the multi-granularity information is fused and processed by using 3D convolution, so that the relevant information in the comment under multiple granularities is effectively highlighted.
(3) The solution is not only superior in performance but also strongly interpretable.
Advantages of additional aspects of the disclosure will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the disclosure.
Drawings
The accompanying drawings, which are included to provide a further understanding of the disclosure, illustrate embodiments of the disclosure and together with the description serve to explain the disclosure and are not to limit the disclosure.
Fig. 1 is a schematic structural diagram of a fine-grained feature interaction network model according to a first embodiment of the present disclosure;
fig. 2 is a diagram illustrating a hierarchical dilation convolution according to a first embodiment of the disclosure.
Detailed Description
The present disclosure is further described with reference to the following drawings and examples.
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present disclosure. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
The embodiments and features of the embodiments in the present disclosure may be combined with each other without conflict.
The first embodiment is as follows:
the embodiment aims to provide a fine-grained item recommendation method based on comment texts.
A fine-grained item recommendation method based on comment texts comprises the following steps:
respectively acquiring a user comment text set and an article comment text set;
calculating a multi-granularity incidence matrix between a user and an article by using a fine-granularity feature interaction network to obtain a 3D interaction image;
inputting the 3D interactive image into a fully-connected neural network and a traditional factorization machine to realize the grading prediction of a user-article, and realizing the recommendation of the article according to the grading result;
the fine-grained feature interaction network specifically includes: performing multi-granularity semantic feature extraction on each comment text in the user comment text set and the article comment text set by using a hierarchical expansion convolution model to obtain a multi-granularity comment expression set of the user and the article; and calculating a multi-granularity incidence matrix between the user and the article, and fusing the incidence matrix into a 3D interactive image through a 3D convolutional neural network.
Specifically, for ease of understanding, the embodiments of the present disclosure are described in detail below with reference to the accompanying drawings:
fig. 1 shows a schematic diagram of a fine-grained feature interactive network model structure, which in turn describes the fine-grained feature interactive network (FFIN) proposed by us. First, we define a recommendation problem based on comments; subsequently, we describe in detail the encoding process of the hierarchical expansion convolution structure in the comment representation module on the original comment text; secondly, the interactive process of the multi-granularity comment representation characteristics of the user and the article is shown; finally, we introduce a scoring prediction layer with a Factorizer (FM) to achieve user scoring of items, where:
(one) problem definition
A comment-based recommendation question may be expressed as: for a given user u and item i, they both correspond to an original set of text sets of commentsAndthe number of comments of the user and item comment sets is N, M, wherein the user comment text sets store comment texts of specific users about different items; and the item comment text set stores comment texts of different users about the specific item. Our goal is to build a predictive model that predicts the user's score for an item through a set of user and item reviews. Assuming that the maximum lengths of the individual comments in a set of user and item comment sets are Tu and Ti, respectively, the original comment texts of the nth user and the mth item may be represented as:
(II) comment text representation
Since the generation process of multi-granular review representations of users and items is similar, we only show the encoding step of a single review text, and willTo simplify to s, the comment length is set to T.
Inspired by the multi-layer extended convolution structure, as shown in fig. 2, we designed a hierarchical extended convolution (HDC) to learn multiple semantic granularity comment representations directly from the comment.
For a given comment text s = [ w ] 1 ,w 2 ,...,w T ]The model is first mapped to the corresponding word vector matrix by an embedding matrix, i.e. X = [ X ]) 1 ,x 2 ,...,x T ]Wherein x is t ∈R d Is the word-embedded representation of the t-th word in the text, and d is the dimension of the word vector. The HDC then takes the word vector matrix as input, capturing the multi-granular semantic features in the comments.
Unlike standard convolution, which convolves a continuous subsequence of inputs at each step, the dilation convolution has a wider acceptance domain by skipping δ input elements at a time, where δ is the dilation rate. For context-centric word x t And a convolution kernel W of size 2w +1, the dilation convolution operation may be formulated as:
wherein,is vector concatenation, b is a bias term, reLU is a non-linear activation function. As shown in fig. 2, the output of each convolutional layer is a weighted combination of the inputs of the previous layer. We start with δ =1 (equivalent to standard convolution) to ensure that no element in the original input sequence is missing. Then, by overlapping the convolutions after expansion with a larger expansion rate level, the length of the convolution text is extended in an exponential manner, and the semantic features of different receiving domains can be covered only by using a small number of layers and a proper amount of parameters.
Furthermore, to prevent the disappearance of the gradient or explosion, we apply layer normalization (layer normalization) at the end of each convolutional layer. Since irrelevant information may be introduced into a long-distance semantic unit, we actually design a multi-level expansion rate according to the performance in verification. The output of each stack layer l is saved as a feature map of the text at a particular level of granularity, formulated asWhere f is the number of convolution filters per layer. Given the dilated convolution of the L-layer stack, the multi-granular comment representation can be defined as [ s ] 1 ,s 2 ,...,s L ]. In this way, the HDC gradually harvests word sense and semantic features from the word and phrase level at a small expansion rate, capturing long-term dependencies from the sentence level at a large expansion rate. Our review coding module not only outperforms recurrent neural networks in terms of parallelism capability, but also significantly reduces memory consumption compared to a fully attention-based approach.
Through HDC module encoding, the multi-granular review representation of a single user and item can be expressed as:
wherein,set of comments S for user and item, respectively u 、S i The nth and mth comment texts in (1).
(III) feature interaction Module
Multi-granular set of comment representations for a given user and itemAndwe achieve feature interaction between users and items by building a correlation matrix at each granularity. Given the nth and mth multi-granular review representations of users and items, the correlation matrix of the l-th layer granularity may be implemented by a dot product operation as follows:
wherein L = [1,2]Is the number of semantic granularities that the user may desire, element (1) ofRepresenting the relevance of the ith feature and the jth feature in the user and item reviews under the granularity of l.
To summarize the multi-granularity features in the reviews of the entire user and item, the FFIN fuses the incidence matrices obtained from all interactions of the user and item into one 3D interactive image, with the formula:
wherein n × m represents the total number of the incidence matrices under a single granularity, and each pixel point in the 3D interactive image is obtained by:
specifically, each pixel point is a spliced vector of features at all granularities, and represents the degree of interaction between a user and an article at multiple granularities.
Since the scoring behavior of the user is usually personalized with subjectivity, different preferences are shown when different articles are faced. We identify a prominent matching signal from the entire image using a hierarchical 3D convolutional neural network and max pooling operations. Three-dimensional convolution is an extension of typical two-dimensional convolution, with both the filter and the step size being three-dimensional cubes. Formally, the higher order pixel on the z-th feature map of the t-th layer is calculated at (k, i, j) by:
wherein ELU is a nonlinear activation function, z' represents a characteristic diagram of a previous layer,and b (t) A 3D convolution kernel and an offset term, respectively, of size W t ×H t ×R t . Then, significant information is extracted by a max pooling operation:
wherein,andis the size of the 3D pooling operation. The output of the last layer is to splice the comprehensive interaction vectors between the users and the article comments, and the vectors are expressed as
(IV) score prediction module
In this subsection, we use the prediction vectors generated by the feature interaction module described aboveInput into a fully connected neural network and a conventional Factoring Machine (FM) to achieve a final user-item score prediction. The implementation of a fully-connected neural network can be formulated as follows:
wherein, W 0 ∈R n×e And b 0 ∈R n Respectively, a weight parameter and a bias term for a fully connected network.
FM accepts a real-valued feature vector and models the interaction between features using decomposition parameters. It is defined as follows
Wherein,is a real-valued input feature vector.<.,.>Is a dot product operation. Parameter { v } 1 ,...,v n },υ∈R n Is for modeling an interaction pair (x) i ,x i ) The decomposition parameter of (2).Andrespectively, a global bias term and a linear regression component. Weight parameter of FM. Output of FMIs a scalar quantity representing the predicted scores of user and item interactions.
Further, different articles are ranked from large to small according to the prediction scores of the articles, and the articles with high prediction scores are recommended to the user.
Example two:
the embodiment aims to provide a fine-grained item recommendation system based on comment texts.
A comment text-based fine-grained item recommendation system comprising:
the data acquisition unit is used for respectively acquiring a user comment text set and an article comment text set;
the multi-granularity feature extraction unit is used for respectively extracting multi-granularity semantic features of each comment text in the user comment text set and the article comment text set by utilizing the hierarchical expansion convolution model to obtain multi-granularity comment expression sets of the user and the article;
the interactive image generation unit is used for constructing a multi-granularity incidence matrix between the user and the article based on the multi-granularity comment representation of the user and the article, and fusing the incidence matrix into a 3D interactive image through a 3D convolution neural network;
and the item recommending unit is used for inputting the 3D interactive image into the fully-connected neural network and the traditional factorization machine to realize the grading prediction of the user-item and realize the recommendation of the item according to the grading result.
In further embodiments, there is also provided:
an electronic device comprising a memory and a processor, and computer instructions stored on the memory and executed on the processor, the computer instructions when executed by the processor performing the method of embodiment one. For brevity, no further description is provided herein.
It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general purpose processors, digital signal processors DSP, application specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and so on. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory may include both read-only memory and random access memory, and may provide instructions and data to the processor, and a portion of the memory may also include non-volatile random access memory. For example, the memory may also store device type information.
A computer readable storage medium storing computer instructions which, when executed by a processor, perform the method of embodiment one.
The method in the first embodiment may be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules in the processor. The software modules may be located in ram, flash, rom, prom, or eprom, registers, among other storage media as is well known in the art. The storage medium is located in a memory, and a processor reads information in the memory and combines hardware thereof to complete the steps of the method. To avoid repetition, it is not described in detail here.
Those of ordinary skill in the art will appreciate that the various illustrative elements, i.e., algorithm steps, described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
The fine-grained article recommendation method and system based on the comment text can be realized, and have a wide application prospect.
The above description is only a preferred embodiment of the present disclosure and is not intended to limit the present disclosure, and various modifications and changes may be made to the present disclosure by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present disclosure should be included in the protection scope of the present disclosure.
Although the present disclosure has been described with reference to specific embodiments, it should be understood that the scope of the present disclosure is not limited thereto, and those skilled in the art will appreciate that various modifications and changes can be made without departing from the spirit and scope of the present disclosure.
Claims (10)
1. A fine-grained item recommendation method based on comment texts is characterized by comprising the following steps:
respectively acquiring a user comment text set and an article comment text set; the user comment text set and the article comment text set respectively correspond to a group of comment texts;
calculating a multi-granularity incidence matrix between a user and an article by using a fine-granularity feature interaction network to obtain a 3D interaction image;
inputting the 3D interactive image into a fully-connected neural network and a traditional factorization machine to realize the grading prediction of a user-article, and realizing the recommendation of the article according to the grading result;
the fine-grained feature interaction network specifically includes: performing multi-granularity semantic feature extraction on each comment text in the user comment text set and the article comment text set respectively by using a hierarchical expansion convolution model to obtain multi-granularity comment expression sets of the user and the article; and calculating a multi-granularity incidence matrix between the user and the object, and fusing the incidence matrix into a 3D interactive image through a 3D convolutional neural network.
2. A comment text based fine grain item recommendation method according to claim 1, wherein the use of the hierarchical dilation convolution model to convolve a continuous subsequence of inputs at each step differently from the standard convolution, the different levels of dilation convolution operate by skipping δ input elements at a time, and the dilation operation is specifically described as:
wherein,is vector concatenation, b is a bias term, reLU is a non-linear activation function, x t Representing the context center word, W is a convolution kernel of size 2w +1, W is a parameter for adjusting the size of the convolution kernel, and δ is an expansion rate, which is an integer not less than 1.
3. A comment text based fine grain item recommendation method as claimed in claim 2 wherein to prevent the disappearance or explosion of the gradient, the last application layer of the dilation convolution at each layer is normalized.
4. The method for recommending fine-grained items based on comment texts as claimed in claim 1, wherein said calculating a multi-grained association matrix between the user and the item specifically comprises: the interaction of features between users and articles is realized by constructing a correlation matrix on each granularity, and given the nth and mth multi-granularity comment representations of the users and the articles, the correlation matrix of the ith layer of granularity is realized by the following dot product operation:
5. The fine-grained item recommendation method based on comment text according to claim 1, wherein the correlation matrix is fused into a 3D interactive image through a 3D convolutional neural network, wherein each pixel point in the 3D interactive image is a spliced vector of features at all granularities, and represents the degree of interaction between a user and an item at multiple granularities.
6. The comment text-based fine-grained item recommendation method according to claim 1, wherein the 3D interactive image is input into a fully-connected neural network and a traditional factorization machine to realize user-item rating prediction, and specifically comprises the following steps: inputting the 3D interactive image into a full-connection neural network to obtain a real-value feature vector, inputting the feature vector into a factor decomposition machine, modeling the interaction between features by using decomposition parameters, and outputting a prediction score of the interaction between the user and the article.
7. The comment text-based fine-grained item recommendation method according to claim 1, wherein before multi-grained semantic feature extraction is performed on comment texts by adopting a hierarchical expansion convolution model, each comment text is mapped into a corresponding word vector matrix through an embedding matrix.
8. A comment text-based fine-grained item recommendation system is characterized by comprising:
the data acquisition unit is used for respectively acquiring a user comment text set and an article comment text set; the user comment text set and the article comment text set respectively correspond to a group of comment texts;
the multi-granularity feature extraction unit is used for respectively extracting multi-granularity semantic features of each comment text in the user comment text set and the article comment text set by utilizing the hierarchical expansion convolution model to obtain multi-granularity comment expression sets of the user and the article;
the interactive image generation unit is used for constructing a multi-granularity incidence matrix between the user and the article based on the multi-granularity comment expression of the user and the article, and fusing the incidence matrix into a 3D interactive image through a 3D convolutional neural network;
and the item recommending unit is used for inputting the 3D interactive image into the fully-connected neural network and the traditional factorization machine to realize the grading prediction of the user-item and realize the recommendation of the item according to the grading result.
9. An electronic device comprising a memory, a processor, and a computer program stored and executed on the memory, wherein the processor when executing the program implements a comment text based fine grain item recommendation system as recited in any one of claims 1 to 7.
10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the program when executed by a processor implements a comment text based fine grain item recommendation system as recited in any one of claims 1 to 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110505981.5A CN113127604B (en) | 2021-05-10 | 2021-05-10 | Comment text-based fine-grained item recommendation method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110505981.5A CN113127604B (en) | 2021-05-10 | 2021-05-10 | Comment text-based fine-grained item recommendation method and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113127604A CN113127604A (en) | 2021-07-16 |
CN113127604B true CN113127604B (en) | 2022-10-04 |
Family
ID=76781246
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110505981.5A Active CN113127604B (en) | 2021-05-10 | 2021-05-10 | Comment text-based fine-grained item recommendation method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113127604B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113570154B (en) * | 2021-08-09 | 2024-07-05 | 齐鲁工业大学 | Multi-granularity interaction recommendation method and system integrating dynamic interests of users |
CN114154071B (en) * | 2021-12-09 | 2023-05-09 | 电子科技大学 | Emotion time sequence recommendation method based on attention mechanism |
CN114676315A (en) * | 2022-01-28 | 2022-06-28 | 齐鲁工业大学 | Method and system for constructing attribute fusion interaction recommendation model based on enhanced graph convolution |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109740154A (en) * | 2018-12-26 | 2019-05-10 | 西安电子科技大学 | A kind of online comment fine granularity sentiment analysis method based on multi-task learning |
CN110060132A (en) * | 2019-04-24 | 2019-07-26 | 吉林大学 | Interpretable Method of Commodity Recommendation based on fine-grained data |
CN111061951A (en) * | 2019-12-11 | 2020-04-24 | 华东师范大学 | Recommendation model based on double-layer self-attention comment modeling |
CN112182156A (en) * | 2020-09-28 | 2021-01-05 | 齐鲁工业大学 | Aspect-level interpretable deep network scoring prediction recommendation method based on text processing |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130218914A1 (en) * | 2012-02-20 | 2013-08-22 | Xerox Corporation | System and method for providing recommendations based on information extracted from reviewers' comments |
US9400779B2 (en) * | 2013-06-06 | 2016-07-26 | Xerox Corporation | Method and system for classifying reviewers' comments and recommending related actions in idea-generating social media platforms |
CN107038609A (en) * | 2017-04-24 | 2017-08-11 | 广州华企联信息科技有限公司 | A kind of Method of Commodity Recommendation and system based on deep learning |
CN110083684B (en) * | 2019-04-24 | 2021-11-19 | 吉林大学 | Interpretable recommendation model for fine-grained emotion |
CN112100485B (en) * | 2020-08-20 | 2024-07-05 | 齐鲁工业大学 | Comment-based scoring prediction article recommendation method and system |
CN112699215B (en) * | 2020-12-24 | 2022-07-05 | 齐鲁工业大学 | Grading prediction method and system based on capsule network and interactive attention mechanism |
-
2021
- 2021-05-10 CN CN202110505981.5A patent/CN113127604B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109740154A (en) * | 2018-12-26 | 2019-05-10 | 西安电子科技大学 | A kind of online comment fine granularity sentiment analysis method based on multi-task learning |
CN110060132A (en) * | 2019-04-24 | 2019-07-26 | 吉林大学 | Interpretable Method of Commodity Recommendation based on fine-grained data |
CN111061951A (en) * | 2019-12-11 | 2020-04-24 | 华东师范大学 | Recommendation model based on double-layer self-attention comment modeling |
CN112182156A (en) * | 2020-09-28 | 2021-01-05 | 齐鲁工业大学 | Aspect-level interpretable deep network scoring prediction recommendation method based on text processing |
Also Published As
Publication number | Publication date |
---|---|
CN113127604A (en) | 2021-07-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US12039280B2 (en) | Multi-turn dialogue response generation with persona modeling | |
CN111611377B (en) | Knowledge distillation-based multi-layer neural network language model training method and device | |
CN113127604B (en) | Comment text-based fine-grained item recommendation method and system | |
US11941366B2 (en) | Context-based multi-turn dialogue method and storage medium | |
EP3893163A1 (en) | End-to-end graph convolution network | |
CN108664589B (en) | Text information extraction method, device, system and medium based on domain self-adaptation | |
CN111881291A (en) | Text emotion classification method and system | |
CN109858041B (en) | Named entity recognition method combining semi-supervised learning with user-defined dictionary | |
US11636341B2 (en) | Processing sequential interaction data | |
CN110598206A (en) | Text semantic recognition method and device, computer equipment and storage medium | |
CN112749274B (en) | Chinese text classification method based on attention mechanism and interference word deletion | |
CN114547298B (en) | Biomedical relation extraction method, device and medium based on combination of multi-head attention and graph convolution network and R-Drop mechanism | |
CN109189862A (en) | A kind of construction of knowledge base method towards scientific and technological information analysis | |
CN110362797B (en) | Research report generation method and related equipment | |
CN113779994B (en) | Element extraction method, element extraction device, computer equipment and storage medium | |
CN110688834A (en) | Method and equipment for rewriting intelligent manuscript style based on deep learning model | |
CN110968725B (en) | Image content description information generation method, electronic device and storage medium | |
CN111145914B (en) | Method and device for determining text entity of lung cancer clinical disease seed bank | |
US20230169271A1 (en) | System and methods for neural topic modeling using topic attention networks | |
Chen et al. | Deep neural networks for multi-class sentiment classification | |
CN112417155A (en) | Court trial query generation method, device and medium based on pointer-generation Seq2Seq model | |
CN112632296A (en) | Knowledge graph-based paper recommendation method and system with interpretability and terminal | |
CN110020431A (en) | Feature extracting method, device, computer equipment and the storage medium of text information | |
CN111581386A (en) | Construction method, device, equipment and medium of multi-output text classification model | |
Cvejoski et al. | Recurrent point review models |
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 | ||
CP03 | Change of name, title or address |
Address after: 250353 University Road, Changqing District, Ji'nan, Shandong Province, No. 3501 Patentee after: Qilu University of Technology (Shandong Academy of Sciences) Country or region after: China Address before: 250353 University Road, Changqing District, Ji'nan, Shandong Province, No. 3501 Patentee before: Qilu University of Technology Country or region before: China |