CN116796045B - Multi-dimensional book grading method, system and readable medium - Google Patents
Multi-dimensional book grading method, system and readable medium Download PDFInfo
- Publication number
- CN116796045B CN116796045B CN202311061921.4A CN202311061921A CN116796045B CN 116796045 B CN116796045 B CN 116796045B CN 202311061921 A CN202311061921 A CN 202311061921A CN 116796045 B CN116796045 B CN 116796045B
- Authority
- CN
- China
- Prior art keywords
- book
- feature vector
- classified
- semantic
- reader
- 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 42
- 239000013598 vector Substances 0.000 claims description 186
- 230000004927 fusion Effects 0.000 claims description 33
- 230000003993 interaction Effects 0.000 claims description 24
- 238000009826 distribution Methods 0.000 claims description 13
- 238000005070 sampling Methods 0.000 claims description 12
- 230000007246 mechanism Effects 0.000 claims description 11
- 238000004590 computer program Methods 0.000 claims description 10
- 230000003044 adaptive effect Effects 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 6
- 230000011218 segmentation Effects 0.000 claims description 6
- 238000010586 diagram Methods 0.000 description 11
- 238000013527 convolutional neural network Methods 0.000 description 10
- 230000014509 gene expression Effects 0.000 description 8
- 230000006870 function Effects 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 238000012549 training Methods 0.000 description 5
- 238000007792 addition Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000002996 emotional effect Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000001537 neural effect Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000008451 emotion Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 238000005215 recombination Methods 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
A multi-dimensional book grading method, system and readable medium, it obtains the history reading information of reader; acquiring book topics and book abstracts of a first book to be classified; and determining a reading difficulty level label of the first book to be classified for the reader based on the historical reading information of the reader and the book theme and the book abstract of the first book to be classified. The reading difficulty level obtained in this way has specificity and self-adaptability to the reading capability of a specific reader, so that a proper book is conveniently recommended for the specific reader.
Description
Technical Field
The application relates to the technical field of intelligent book classification, and in particular relates to a multi-dimensional book classification method, a system and a readable medium.
Background
The reading capability of each reader is different, and the receiving capability of each reader on books to be read is also different, so that the reading requirements of different readers are also different, namely, the different readers are required to be classified in a targeted way for reading the books.
In the traditional book grading system, books are rated hard through artificially set grading rules, for example, grading is carried out through age stages, and the grading is mass-oriented grading, so that universality is achieved, and reading experience and reading requirements of each reader cannot be well met.
Thus, an optimized book ranking scheme is desired.
Disclosure of Invention
The present application has been made to solve the above-mentioned technical problems. The embodiment of the application provides a multi-dimensional book grading method, a multi-dimensional book grading system and a readable medium, wherein the multi-dimensional book grading system acquires historical reading information of readers; acquiring book topics and book abstracts of a first book to be classified; and determining a reading difficulty level label of the first book to be classified for the reader based on the historical reading information of the reader and the book theme and the book abstract of the first book to be classified. The reading difficulty level obtained in this way has specificity and self-adaptability to the reading capability of a specific reader, so that a proper book is conveniently recommended for the specific reader.
In a first aspect, a multi-dimensional book grading method is provided, comprising:
Acquiring historical reading information of a reader;
acquiring book topics and book abstracts of a first book to be classified; and
and determining the reading difficulty level label of the first book to be classified for the reader based on the historical reading information of the reader and the book theme and the book abstract of the first book to be classified.
In a second aspect, there is provided a multi-dimensional book ranking system comprising:
the reading information acquisition module is used for acquiring historical reading information of the reader;
the theme and abstract acquisition module is used for acquiring book theme and book abstract of the first book to be classified; and
the difficulty level label determining module is used for determining the reading difficulty level label of the first book to be classified for the reader based on the historical reading information of the reader, the book theme and the book abstract of the first book to be classified.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a multi-dimensional book ranking method according to an embodiment of the application.
Fig. 2 is a schematic diagram of a multi-dimensional book grading method according to an embodiment of the application.
FIG. 3 is a flowchart of the sub-steps of step 130 in a multi-dimensional book ranking method according to an embodiment of the application.
FIG. 4 is a block diagram of a multi-dimensional book ranking system according to an embodiment of the application.
Fig. 5 is a schematic view of a scenario of a multi-dimensional book grading method according to an embodiment of the application.
Detailed Description
The following description of the technical solutions according to the embodiments of the present application will be given with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
Unless defined otherwise, all technical and scientific terms used in the embodiments of the application have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the present application is for the purpose of describing particular embodiments only and is not intended to limit the scope of the present application.
In describing embodiments of the present application, unless otherwise indicated and limited thereto, the term "connected" should be construed broadly, for example, it may be an electrical connection, or may be a communication between two elements, or may be a direct connection, or may be an indirect connection via an intermediate medium, and it will be understood by those skilled in the art that the specific meaning of the term may be interpreted according to circumstances.
It should be noted that, the term "first\second\third" related to the embodiment of the present application is merely to distinguish similar objects, and does not represent a specific order for the objects, it is to be understood that "first\second\third" may interchange a specific order or sequence where allowed. It is to be understood that the "first\second\third" distinguishing objects may be interchanged where appropriate such that embodiments of the application described herein may be practiced in sequences other than those illustrated or described herein.
It should be appreciated that the conventional book rating system is a method of rating books by artificially set rating rules. Such a rating system is typically conducted on an age basis, with the aim of providing readers of different ages with books that suit their reading capabilities and content interests.
Conventional book rating systems are generally formulated by publishers, educational institutions or related expert organizations and are widely used in the field of book markets and education, and they divide books into different levels or age groups according to the contents of the books, the difficulty of language, the complexity of the episodes, etc.
In conventional book rating systems, symbols like letters, numbers or colors are often used to represent the different ratings, for example, common rating symbols include "Ages 4-8" (4-8 years old apply), "Young add" (teenager apply), and the like. These symbols are typically printed on the front, back, or catalog pages of the book so that readers and purchasers can quickly learn about the applicable objects of the book.
The advantage of conventional book-grading systems is their versatility and easy understanding, providing the reader with a reference to help them select the appropriate reading material among the numerous books. At the same time, the grading system is also helpful for parents, teachers and library administrators to better select proper books for children and teenagers.
However, conventional book rating systems also have some limitations. Since the ranking is based on a generalized age group or stage, it may not meet individual differences and special needs of each reader. Some readers may have a higher reading power at a certain age group, while some readers may have a lower reading power at the same age group. In addition, the content and theme of books may also vary depending on personal interests and the level of development, and conventional rating systems cannot fully take these factors into account.
Therefore, in order to better meet the reading experience and requirements of different readers, it may be considered to employ an optimized book ranking scheme based on personalized reading capabilities to more accurately evaluate the level of the reader's capabilities and provide them with more suitable reading materials according to individual differences.
In the application, a multi-dimensional book classification combined with convolutional neural networks (Convolutional Neural Network, CNN) is provided, which is a method for improving the accuracy and individuation of book classification. Conventional book rating systems are typically classified based on age group or reading ability, but this approach ignores other important features of the book and individual differences of the reader.
Further, conventional book ranking systems focus mainly on the age and reading ability of readers, while ignoring other features of books, such as theme, scenario complexity, emotional tendency, etc. Combining with CNN can evaluate the difficulty of the book and the characteristics of the reader more comprehensively by analyzing the text content and other metadata of the book and considering a plurality of features at the same time.
CNN is a deep learning model with strong feature extraction capability. By training the CNN model, key features in book texts, such as vocabulary, syntax structure, theme and the like, can be automatically learned and extracted. These features can be used to evaluate the difficulty level of the book and match the personalized model of the reader to more accurately determine if the book is suitable for the reader.
The personalized rating result can be realized by combining the multi-dimensional book grading of the CNN, and the individual difference and preference of the readers can be taken into consideration by establishing a personalized model of the readers and training the CNN model, so that book recommendation suitable for the reading capability, interest and preference of each reader is provided for each reader. The personalized grading method can improve the reading experience of readers and better meet the reading requirements of the readers.
The CNN model has a certain interpretability, and the basis of the model for grading books can be understood through a visualization technology. This provides transparency to the book ranking results, allowing readers and related experts to understand how the model is evaluated and recommended, and to verify and adjust the results.
That is, the multi-dimensional book classification combined with the convolutional neural network can improve the classification accuracy, individuation degree and transparency, consider a plurality of characteristics of books and individual differences of readers, provide more accurate and individuated book recommendation for readers, improve reading experience and meet reading requirements.
FIG. 1 is a flow chart of a multi-dimensional book ranking method according to an embodiment of the application. Fig. 2 is a schematic diagram of a multi-dimensional book grading method according to an embodiment of the application. As shown in fig. 1 and 2, the multi-dimensional book grading method includes: 110, acquiring historical reading information of readers; 120, obtaining book topics and book abstracts of a first book to be classified; and 130, determining a reading difficulty level label of the first book to be classified for the reader based on the historical reading information of the reader and the book theme and the book abstract of the first book to be classified.
Wherein, in the step 110, it is ensured that enough historical reading information is obtained so as to accurately evaluate the reading capability and preference of the reader, which may include information on the books read, the reading time length, the reading speed, and the like. By analyzing the historical reading information of the reader, the reading level, the hobbies and the reading preference of the reader can be known. The suitability and difficulty of books can be evaluated in a personalized way, and more accurate book recommendation can be provided for readers.
In the step 120, accurate and comprehensive book theme and summary information is ensured to be acquired, which can be acquired through channels of metadata of books, information provided by publishers, comments of books and the like. The theme and abstract of the book provide important clues about the content and complexity of the book, and by analyzing this information, the characteristics of the book, including difficulty, plot complexity, emotional tendency, etc., can be better understood, thereby grading more accurately.
In the step 130, when the historical reading information of the reader and the information such as the subject matter, abstract and the like of the book to be classified are combined, a proper model or algorithm is established for evaluation and classification. This may involve techniques of feature extraction, machine learning, deep learning, etc. By comprehensively considering individual differences of readers and characteristics of books to be classified, the proper difficulty level label of the books can be determined. This helps to provide personalized book recommendations for readers, improving book classification accuracy and adaptability.
In the above steps, the multi-dimensional book grading method involves obtaining historical reading information of readers, obtaining theme and summary information of books to be graded, and determining reading difficulty level labels of the books based on the information. Points of attention include ensuring accuracy and comprehensiveness of the information, building appropriate models or algorithms for evaluation and classification. The beneficial effects of these methods include providing personalized book recommendations, improving the accuracy and adaptability of the ranking.
Specifically, in the step 110 and the step 120, history reading information of the reader is obtained; and acquiring book topics and book abstracts of the first books to be classified. Aiming at the technical problems, the technical concept of the application is to measure the reading capability of a reader based on the historical reading information of the reader, and further to classify the reading difficulty level of the data to be read based on the reading capability, so that the obtained reading difficulty level has the specificity and the self-adaptability aiming at the reading capability of a specific reader, thereby being convenient for recommending a proper book for the specific reader.
Specifically, in the technical scheme of the application, historical reading information of readers is obtained, and book subjects and book summaries of the first books to be classified are obtained. Wherein, through analyzing the historical reading information of the reader, the reading level, reading speed, understanding ability and the like of the reader can be known, and the information is very important for evaluating the reading difficulty and suitability of the reader. Through the historical reading information of the reader, the reading preference, favorite book types, topics and the like of the reader can be known, the suitability of books can be evaluated in a personalized way, and book recommendation meeting the interests of the reader can be provided.
Further, the book's theme and abstract provide important information about the book's content and theme. By analyzing the information, characteristics of the books such as plot complexity, theme type, emotion tendency and the like can be known, so that the difficulty and suitability of the books can be evaluated more accurately. The topic and abstract information of the book can also be used to extract text features of the book, such as vocabulary, syntactic structures, etc., which are helpful in assessing the difficulty and suitability of the book.
By combining historical reading information of readers and information such as themes, abstracts and the like of books to be classified, the proper difficulty level label of the books can be determined, personalized book recommendation is provided for the readers, and the reading capability and interest of the books and the readers are ensured to be matched. By comprehensively considering individual differences of readers and characteristics of books to be classified, the reading difficulty level of the books can be more accurately determined, the accuracy of book classification is improved, the readers can select books suitable for the readers, and reading experience and understanding capability are improved.
The method comprises the steps of obtaining historical reading information of readers and book topics and book abstracts of a first book to be classified, and playing a key role in determining reading difficulty level labels of the books to be classified for the readers. The information is helpful for knowing the reading capability, preference and interest of readers, and provides important clues for book contents and features, so that personalized book recommendation is realized and grading accuracy is improved.
Specifically, in the step 130, based on the historical reading information of the reader and the book theme and the book abstract of the first book to be classified, a reading difficulty level label of the first book to be classified for the reader is determined. FIG. 3 is a flowchart illustrating the substep of step 130 in the multi-dimensional book grading method according to an embodiment of the present application, as shown in FIG. 3, for determining a reading difficulty level tag of the first book to be graded for the reader based on the historical reading information of the reader and the book subject and the book abstract of the first book to be graded, including: 131, carrying out semantic coding on the historical reading information of the reader and the book theme and the book abstract of the first book to be classified so as to obtain a semantic understanding feature vector of the historical reading information and a semantic coding feature vector of the first book to be classified; 132, performing feature interaction on the semantic understanding feature vector of the historical reading information and the semantic coding feature vector of the first book to be classified to obtain a self-adaptive coding feature vector of the first book reading difficulty; and 133, determining a reading difficulty level label of the first book to be classified for the reader based on the self-adaptive coding feature vector of the reading difficulty of the first book.
Firstly, converting historical reading information of readers, topics of a first book to be ranked, abstracts and other text information into semantic representations, including Word embedding (such as Word2Vec and GloVe) and pre-trained language models (such as BERT and GPT). The text information is converted into a numerical feature vector through semantic coding so as to facilitate subsequent feature interaction and model training. These feature vectors capture the semantic and contextual information of the text.
Then, the semantic understanding feature vector of the historical reading information and the semantic encoding feature vector of the first book to be classified are subjected to feature interaction, and various methods such as splicing, weighted addition, attention mechanism and the like can be adopted. Therefore, the information of different features can be fused, and the expression capability and the discrimination capability of the features are improved. The self-adaptive coding feature vector of the reading difficulty of the first book is obtained through feature interaction, historical reading information of readers and semantic features of the first book to be classified are comprehensively considered, and the proper difficulty degree of the book is reflected.
Then, a classification model or regression model can be established to predict the reading difficulty level label of the first book to be classified for the reader by utilizing the self-adaptive coding feature vector of the reading difficulty of the first book, and the label can be discrete difficulty level (such as primary level, medium level and high level) or continuous difficulty scoring. By determining the reading difficulty level label of the first book to be classified, personalized book recommendation can be provided for readers, the reading capability and interest matching of the books and the readers are ensured, and the reading experience and understanding capability are improved.
Through the steps, based on the historical reading information of the reader and the theme and abstract of the first book to be classified, the reading difficulty self-adaptive coding feature vector of the first book can be obtained through semantic coding and feature interaction, and the reading difficulty grade label of the book is determined based on the feature vector. These steps help to achieve personalized book recommendation and improve the accuracy of the classification, providing book selection meeting the reader's needs.
For said step 131, it comprises: after word segmentation is carried out on the historical reading information of the reader, semantic understanding feature vectors of the historical reading information are obtained through a semantic encoder comprising a word embedding layer; and performing word segmentation on the book theme and the book abstract of the first book to be classified, and then obtaining the semantic coding feature vector of the first book to be classified through the semantic coder containing the word embedding layer.
Wherein the semantic encoder is a converter-based Bert model.
And then, carrying out semantic coding on the historical reading information of the reader and the book theme and the book abstract of the first book to be classified so as to obtain a semantic understanding feature vector of the historical reading information and a semantic coding feature vector of the first book to be classified. The method comprises the steps of obtaining historical reading information of a reader, and book topics and book abstracts of a first book to be classified, and then carrying out semantic understanding on the historical reading information of the reader, the book topics and the book abstracts of the first book to be classified so as to obtain historical reading information of the reader, semantic feature representations of the book topics and the book abstracts of the first book to be classified, namely semantic understanding feature vectors of the historical reading information and semantic coding feature vectors of the first book to be classified.
In a specific example of the present application, the process of semantically encoding the historical reading information of the reader and the book theme and the book abstract of the first book to be classified to obtain the semantic understanding feature vector of the historical reading information and the semantic encoding feature vector of the first book to be classified includes: firstly, performing word segmentation on historical reading information of a reader, and then obtaining a semantic understanding feature vector of the historical reading information through a semantic encoder comprising a word embedding layer; and performing word segmentation on the book theme and the book abstract of the first book to be classified, and then obtaining the semantic coding feature vector of the first book to be classified through the semantic coder containing the word embedding layer. In this particular example, the semantic encoder is a transformer-based Bert model.
For the step 132, it includes: and carrying out feature interaction between the historical reading information semantic understanding feature vector and the first book semantic coding feature vector to be classified based on an attention mechanism by using an inter-feature attention interaction layer so as to obtain a first book reading difficulty self-adaptive coding feature vector.
It should be understood that, in the technical solution of the present application, the semantic understanding feature vector of the historical reading information may reflect the reading capability of the reader, and the semantic encoding feature vector of the first book to be classified may reflect the reading difficulty of the first book to be classified. Based on the above, in the technical scheme of the application, the reading difficulty level label of the first book to be classified for the reader is further determined based on the historical reading information of the reader and the book theme and the book abstract of the first book to be classified.
Specifically, firstly, performing feature interaction on the semantic understanding feature vector of the historical reading information and the semantic encoding feature vector of the first book to be classified to obtain a first book reading difficulty self-adaptive encoding feature vector, namely, performing feature level data interaction on the reading capacity of a reader and the reading difficulty of the first book to be classified in a high-dimensional feature space to obtain a collaborative representation between the two, namely, the first book reading difficulty self-adaptive encoding feature vector.
In a specific technical scheme of the application, an attention interaction layer is used for carrying out attention mechanism-based feature interaction between the historical reading information semantic understanding feature vector and the first book semantic coding feature vector to be classified so as to obtain the first book reading difficulty self-adaptive coding feature vector, and the process can be expressed as follows by a formula:
for the historical reading information semantic understanding feature vector and the first book semantic coding feature vector to be ranked,and->It is transformed into two feature spaces s and t to calculate the degree between them:
wherein,is a learned weight matrix corresponding to the 1 x 1 convolution in fig. 1, i is an index of the output location, j represents an index of all possible locations.
In particular, in the technical solution of the present application, the attention mechanism caused by the attention layer between features is mainly different from the traditional attention mechanism in that: the goal of traditional attention mechanisms is to learn an attention weight matrix, applied to the individual neural nodes of the current layer, giving them greater weight for those important nodes and less weight for those secondary nodes. Because each neural node contains certain characteristic information, the neural network can select information which is more critical to the current task target from a plurality of characteristic information through the operation. The attention mechanism provided by the application is different, and more attention is paid to the dependency relationship among the characteristic information, namely, the characteristic level dependency and interaction relationship between the semantic understanding characteristic vector of the history reading information and the semantic encoding characteristic vector of the first book to be classified.
Further, the inter-feature attention interaction layer is an attention mechanism for modeling the degree of association between different features during feature interaction. And using an inter-feature attention interaction layer to realize feature interaction between the semantic understanding feature vector of the historical reading information and the semantic coding feature vector of the first book to be classified so as to obtain the reading difficulty self-adaptive coding feature vector of the first book.
The inter-feature attention interaction layer is used for adaptively fusing the feature representations of the historical reading information and the feature of the first book to be classified according to the association degree between the two features. By introducing an attention mechanism, features can be weighted according to importance between different features, thereby better capturing relevance and importance between features. Therefore, the expression capability of the features can be improved, and the obtained self-adaptive coding feature vector of the first book reading difficulty can more accurately reflect the proper difficulty degree of the book.
For said step 133, it comprises: carrying out homogeneous Gilbert space metric dense point distribution sampling fusion on the semantic understanding feature vector of the historical reading information and the semantic encoding feature vector of the first book to be classified to obtain a fusion feature vector; fusing the fused feature vector and the first book reading difficulty self-adaptive coding feature vector to obtain a classification feature vector; and the classification feature vector passes through a classifier to obtain a classification result, wherein the classification result is used for representing the reading difficulty grade label of the first book to be classified for the reader.
The historical reading information semantic understanding feature vector and the first book semantic coding feature vector to be classified are fused into a fusion feature vector through homogeneous Gilbert space metric type dense point distribution sampling fusion. Thus, the information of different characteristics can be comprehensively utilized, and the expression capacity and the discrimination capacity of the characteristics are improved.
And fusing the fusion feature vector with the self-adaptive coding feature vector of the first book reading difficulty to obtain a classification feature vector. The fusion mode can adaptively fuse information of different features according to specific tasks and data characteristics, and better captures correlation and importance degree among the features.
And obtaining the classification feature vector by fusing the feature vector and the first book reading difficulty self-adaptive coding feature vector. The feature vector comprehensively considers the historical reading information, the semantic features of the first book to be classified and the reading difficulty self-adaptive coding features, and can evaluate the difficulty of the book and the characteristics of suitable readers more comprehensively.
And training and predicting by using the classification feature vector as input through a classifier to obtain a classification result. The classification feature vector fused with a plurality of features can provide more accurate information, and is beneficial to improving the performance and accuracy of the classifier.
Comprehensive classification feature vectors can be obtained through homogeneous Gilbert space metric type dense point distribution sampling fusion and feature fusion, and the reading difficulty level label for readers is obtained through training and prediction by a classifier. The method can more accurately evaluate the difficulty of the book and the characteristics of the proper readers, provide personalized book recommendation, promote reading experience and meet reading requirements.
In particular, in the technical scheme of the application, because the attention mechanism-based feature interaction between the historical reading information semantic understanding feature vector and the to-be-classified first book semantic coding feature vector is performed by using the attention interaction layer, the dependency relationship between the historical reading information semantic understanding feature vector and the to-be-classified first book semantic coding feature vector is extracted, and for the historical reading information semantic understanding feature vector and the to-be-classified first book semantic coding feature vector, the expression of the historical reading information of the reader and the text semantic features of the book subject and the book abstract of the to-be-classified first book by the historical reading information semantic understanding feature vector and the to-be-classified first book semantic coding feature vector can be insufficient, so that the applicant of the application considers further fusing the text semantic feature expressions of the historical reading information semantic understanding feature vector and the to-be-classified first book semantic coding feature vector to strengthen the feature expressions of the first book reading difficulty adaptive coding feature vector.
Moreover, the applicant considers that the semantic understanding feature vector of the historical reading information and the semantic encoding feature vector of the first book to be classified are respectively the historical reading information of the reader and the closely-encoded dense feature collection type text semantic expression of the book subject and the book abstract of the first book to be classified, so that the semantic understanding feature vector of the historical reading information is recorded as And the first book semantic coding feature vector to be ranked, e.g. denoted +.>Performing homogeneous Gilbert spatial metric dense point distribution sampling fusionThe method is specifically expressed as follows: carrying out homogeneous Gilbert space metric dense point distribution sampling fusion on the semantic understanding feature vector of the historical reading information and the semantic encoding feature vector of the first book to be classified by using the following optimization formula to obtain a fused feature vector; wherein, the optimization formula is:
wherein,represent Min distance and +.>Is super-parameter (herba Cinchi Oleracei)>Is the semantic understanding feature vector of the history reading information, < >>Is the semantic coding feature vector of the first book to be classified,>and->The semantic understanding feature vector of the history reading information is +.>And the first book semantic coding feature vector to be classified +.>Is characterized by the global feature mean value of the history reading information semantic understanding feature vector +.>And the first book semantic coding feature vector to be classified +.>For row vector +.>Is the transpose vector of the semantically encoded feature vector of the first book to be ranked,/or%>And->Representing addition by position and multiplication by position, +.>Is the fusion feature vector.
Here, the feature vector is understood semantically by reading the information on the history And the first book semantic coding feature vector to be classified +.>For semantic understanding of feature vectors for said history read information>And the first book semantic coding feature vector to be classified +.>The fusion feature distribution of the cross-distance constraint point-by-point feature association is used as a bias term to realize feature dense point sampling pattern distribution fusion in the association constraint limit of the feature distribution, so that the homogeneous sampling association fusion among vectors is enhanced. Then, the fusion feature vector is added again>Fusion with the first book reading difficulty self-adaptive coding feature vector can improve the first book reading difficulty self-adaptive coding feature vectorAnd the first book reading difficulty self-adaptively codes the characteristic expression of the characteristic vector.
Further, fusing the fused feature vector and the first book reading difficulty adaptive coding feature vector to obtain a classification feature vector, including: fusing the fusion feature vector and the first book reading difficulty self-adaptive coding feature vector by the following fusion formula to obtain a classification feature vector; wherein, the fusion formula is:
Wherein,representing the fusion feature vector and the first book reading difficulty adaptive coding feature vector,/I>Representing a cascade function->Representing the classification feature vector.
And determining a reading difficulty level label of the first book to be classified for the reader based on the self-adaptive coding feature vector of the reading difficulty of the first book. In a specific example of the application, the first book reading difficulty adaptive coding feature vector is passed through a classifier to obtain a classification result, wherein the classification result is used for representing a reading difficulty level label.
In summary, the multi-dimensional book classification method 100 according to the embodiment of the application is illustrated, measures the reading capability of a reader based on the historical reading information of the reader, and classifies the reading difficulty level of the data to be read based on the reading capability, so that the obtained reading difficulty level has specificity and adaptability to the reading capability of a specific reader, thereby facilitating the recommendation of a proper book for the specific reader.
In one embodiment of the present application, FIG. 4 is a block diagram of a multi-dimensional book ranking system according to an embodiment of the present application. As shown in fig. 4, the multi-dimensional book rating system 200 according to an embodiment of the present application includes: a reading information obtaining module 210, configured to obtain historical reading information of a reader; a theme and abstract obtaining module 220, configured to obtain book theme and book abstract of the first book to be classified; and a difficulty level tag determining module 230, configured to determine a reading difficulty level tag of the first book to be classified for the reader based on the historical reading information of the reader and the book subject and the book abstract of the first book to be classified.
Here, it will be understood by those skilled in the art that the specific functions and operations of the respective units and modules in the above-described multi-dimensional book grading system have been described in detail in the above description of the multi-dimensional book grading method with reference to fig. 1 to 3, and thus, repetitive descriptions thereof will be omitted.
As described above, the multi-dimensional book rating system 200 according to the embodiment of the present application may be implemented in various terminal devices, such as a server for multi-dimensional book rating, etc. In one example, the multi-dimensional book rating system 200 according to embodiments of the present application may be integrated into the terminal device as a software module and/or hardware module. For example, the multi-dimensional book rating system 200 may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the multi-dimensional book rating system 200 may also be one of a number of hardware modules of the terminal device.
Alternatively, in another example, the multi-dimensional book rating system 200 and the terminal device may be separate devices, and the multi-dimensional book rating system 200 may be connected to the terminal device through a wired and/or wireless network and transmit interactive information in a agreed data format.
In one embodiment of the application, a readable medium is provided, characterized in that it has stored thereon computer program instructions, which when executed by a processor, cause the processor to perform a multi-dimensional book ranking method as described.
It should be appreciated that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the forms of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects may be utilized. Furthermore, the computer program product may take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
Methods, systems, and computer program products of embodiments of the present application are described in the flow diagrams and/or block diagrams. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Fig. 5 is a schematic view of a scenario of a multi-dimensional book grading method according to an embodiment of the application. As shown in fig. 5, in the application scenario, first, history reading information of a reader (e.g., C1 as illustrated in fig. 5) and book subjects and book summaries of a first book to be classified (e.g., C2 as illustrated in fig. 5) are acquired; the obtained historical reading information, book theme of the first book and book abstract are then input into a server (e.g., S as illustrated in fig. 5) deployed with a multi-dimensional book ranking algorithm, wherein the server is capable of processing the historical reading information, book theme of the first book and book abstract based on the multi-dimensional book ranking algorithm to determine a reading difficulty level tag of the first book to be ranked for the reader.
It is also noted that in the apparatus, devices and methods of the present application, the components or steps may be disassembled and/or assembled. Such decomposition and/or recombination should be considered as equivalent aspects of the present application.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description has been presented for purposes of illustration and description. Furthermore, this description is not intended to limit embodiments of the application to the form disclosed herein. Although a number of example aspects and embodiments have been discussed above, a person of ordinary skill in the art will recognize certain variations, modifications, alterations, additions, and subcombinations thereof.
Claims (7)
1. A multi-dimensional book grading method, comprising:
acquiring historical reading information of a reader;
Acquiring book topics and book abstracts of a first book to be classified; and
determining a reading difficulty level label of the first book to be classified for the reader based on the historical reading information of the reader and the book theme and the book abstract of the first book to be classified;
based on the historical reading information of the reader and the book theme and the book abstract of the first book to be classified, determining the reading difficulty level label of the first book to be classified for the reader comprises the following steps:
carrying out semantic coding on the historical reading information of the reader and the book theme and the book abstract of the first book to be classified so as to obtain a semantic understanding feature vector of the historical reading information and a semantic coding feature vector of the first book to be classified;
performing feature interaction on the semantic understanding feature vector of the historical reading information and the semantic coding feature vector of the first book to be classified to obtain a self-adaptive coding feature vector of the first book reading difficulty; and
determining a reading difficulty level label of the first book to be classified for the reader based on the self-adaptive coding feature vector of the reading difficulty of the first book;
Based on the self-adaptive coding feature vector of the reading difficulty of the first book, determining the reading difficulty level label of the first book to be classified for the reader comprises the following steps:
carrying out homogeneous Gilbert space metric dense point distribution sampling fusion on the semantic understanding feature vector of the historical reading information and the semantic encoding feature vector of the first book to be classified to obtain a fusion feature vector;
fusing the fused feature vector and the first book reading difficulty self-adaptive coding feature vector to obtain a classification feature vector; and
the classification feature vector passes through a classifier to obtain a classification result, wherein the classification result is used for representing a reading difficulty level label of a first book to be classified aiming at the reader;
carrying out homogeneous Gilbert space metric dense point distribution sampling fusion on the historical reading information semantic understanding feature vector and the first book semantic coding feature vector to be classified to obtain a fusion feature vector, wherein the method comprises the following steps: carrying out homogeneous Gilbert space metric dense point distribution sampling fusion on the semantic understanding feature vector of the historical reading information and the semantic encoding feature vector of the first book to be classified by using the following optimization formula to obtain a fused feature vector;
Wherein, the optimization formula is:
wherein L is p (. Cndot.). Cndot.represents the mintype distance, and p is a superparameter, V 1 Is the semantic understanding feature vector of the history reading information, V 2 Is the semantic coding feature vector of the first book to be classified,and->The semantic understanding feature vector V of the history reading information 1 And the semantic coding feature vector V of the first book to be classified 2 And said history read information semantically understand feature vector V 1 And the semantic coding feature vector V of the first book to be classified 2 As row vector, V 2 T Is the transpose vector of the semantically encoded feature vector of the first book to be ranked,/or%>And "+. c Is the fusion feature vector.
2. The multi-dimensional book classification method according to claim 1, wherein semantically encoding the historical reading information of the reader and the book theme and the book abstract of the first book to be classified to obtain a semantic understanding feature vector of the historical reading information and a semantic encoding feature vector of the first book to be classified, comprises:
after word segmentation is carried out on the historical reading information of the reader, semantic understanding feature vectors of the historical reading information are obtained through a semantic encoder comprising a word embedding layer; and
And performing word segmentation on the book theme and the book abstract of the first book to be classified, and then obtaining the semantic coding feature vector of the first book to be classified through the semantic coder containing the word embedding layer.
3. The multi-dimensional book grading method according to claim 2, wherein the semantic encoder is a converter-based Bert model.
4. The method of claim 3, wherein performing feature interaction on the historical reading information semantic understanding feature vector and the first book semantic coding feature vector to be classified to obtain a first book reading difficulty adaptive coding feature vector comprises:
and carrying out feature interaction between the historical reading information semantic understanding feature vector and the first book semantic coding feature vector to be classified based on an attention mechanism by using an inter-feature attention interaction layer so as to obtain a first book reading difficulty self-adaptive coding feature vector.
5. The multi-dimensional book ranking method of claim 4, wherein fusing the fused feature vector and the first book reading difficulty adaptive coding feature vector to obtain a classification feature vector, comprising: fusing the fusion feature vector and the first book reading difficulty self-adaptive coding feature vector by the following fusion formula to obtain a classification feature vector;
Wherein, the fusion formula is:
V m ＝Concat[V c ，V 1 ]
wherein V is c ，V 1 Representing the fusion feature vector and the self-adaptive coding feature vector of the first book reading difficulty, concat [. Cndot.]Representing a cascade function, V m Representing the classification feature vector.
6. A multi-dimensional book grading system, comprising:
the reading information acquisition module is used for acquiring historical reading information of the reader;
the theme and abstract acquisition module is used for acquiring book theme and book abstract of the first book to be classified; and
the difficulty level label determining module is used for determining the reading difficulty level label of the first book to be classified for the reader based on the historical reading information of the reader and the book theme and the book abstract of the first book to be classified;
the difficulty level label determining module is specifically configured to:
carrying out semantic coding on the historical reading information of the reader and the book theme and the book abstract of the first book to be classified so as to obtain a semantic understanding feature vector of the historical reading information and a semantic coding feature vector of the first book to be classified;
performing feature interaction on the semantic understanding feature vector of the historical reading information and the semantic coding feature vector of the first book to be classified to obtain a self-adaptive coding feature vector of the first book reading difficulty; and
Determining a reading difficulty level label of the first book to be classified for the reader based on the self-adaptive coding feature vector of the reading difficulty of the first book;
the difficulty level label determining module is specifically configured to:
carrying out homogeneous Gilbert space metric dense point distribution sampling fusion on the semantic understanding feature vector of the historical reading information and the semantic encoding feature vector of the first book to be classified to obtain a fusion feature vector;
fusing the fused feature vector and the first book reading difficulty self-adaptive coding feature vector to obtain a classification feature vector; and
the classification feature vector passes through a classifier to obtain a classification result, wherein the classification result is used for representing a reading difficulty level label of a first book to be classified aiming at the reader;
carrying out homogeneous Gilbert space metric dense point distribution sampling fusion on the semantic understanding feature vector of the historical reading information and the semantic encoding feature vector of the first book to be classified by using the following optimization formula to obtain a fused feature vector;
wherein, the optimization formula is:
wherein L is p (. Cndot.). Cndot.represents the mintype distance, and p is a superparameter, V 1 Is the semantic understanding feature vector of the history reading information, V 2 Is the semantic coding feature vector of the first book to be classified,and->The semantic understanding feature vector V of the history reading information 1 And the semantic coding feature vector V of the first book to be classified 2 And said history read information semantically understand feature vector V 1 And the semantic coding feature vector V of the first book to be classified 2 As row vector, V 2 T Is the transpose vector of the semantically encoded feature vector of the first book to be ranked,/or%>And "+. c Is the fusion feature vector.
7. A readable medium having stored thereon computer program instructions which, when executed by a processor, cause the processor to perform the multi-dimensional book grading method according to any of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311061921.4A CN116796045B (en) | 2023-08-23 | 2023-08-23 | Multi-dimensional book grading method, system and readable medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311061921.4A CN116796045B (en) | 2023-08-23 | 2023-08-23 | Multi-dimensional book grading method, system and readable medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116796045A CN116796045A (en) | 2023-09-22 |
| CN116796045B true CN116796045B (en) | 2023-11-10 |
Family
ID=88050053
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311061921.4A Active CN116796045B (en) | 2023-08-23 | 2023-08-23 | Multi-dimensional book grading method, system and readable medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116796045B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117033733B (en) * | 2023-10-09 | 2023-12-22 | 北京民谐文化传播有限公司 | Intelligent automatic classification and label generation system and method for library resources |
| CN117457134A (en) * | 2023-11-06 | 2024-01-26 | 广东信拓人工智能研究院有限公司 | Medical data management method and system based on intelligent AI |
| CN118261117B (en) * | 2024-04-26 | 2024-08-20 | 莱芜职业技术学院 | Book rating intelligent processing method based on big data resource service |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112347239A (en) * | 2020-11-12 | 2021-02-09 | 方正株式(武汉)科技开发有限公司 | Book matching method, system, server and storage medium based on big data |
| CN113590970A (en) * | 2021-08-11 | 2021-11-02 | 东北林业大学 | Personalized digital book recommendation system and method based on reader preference, computer and storage medium |
| CN113610680A (en) * | 2021-08-17 | 2021-11-05 | 山西传世科技有限公司 | AI-based interactive reading material personalized recommendation method and system |
| CN113935332A (en) * | 2021-11-02 | 2022-01-14 | 海信集团控股股份有限公司 | Book grading method and book grading equipment |
| CN114818921A (en) * | 2022-04-26 | 2022-07-29 | 北京思源智通科技有限责任公司 | Multi-dimensional book grading method, system, readable medium and equipment |
| CN115630140A (en) * | 2022-11-02 | 2023-01-20 | 云南师范大学 | English reading material difficulty judgment method based on text feature fusion |
| CN116401466A (en) * | 2023-06-08 | 2023-07-07 | 北京奇趣万物科技有限公司 | Book classification recommendation method and system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR102410253B1 (en) * | 2021-12-21 | 2022-06-22 | (주)웅진씽크빅 | System and method for providing personalized book |
-
2023
- 2023-08-23 CN CN202311061921.4A patent/CN116796045B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112347239A (en) * | 2020-11-12 | 2021-02-09 | 方正株式(武汉)科技开发有限公司 | Book matching method, system, server and storage medium based on big data |
| CN113590970A (en) * | 2021-08-11 | 2021-11-02 | 东北林业大学 | Personalized digital book recommendation system and method based on reader preference, computer and storage medium |
| CN113610680A (en) * | 2021-08-17 | 2021-11-05 | 山西传世科技有限公司 | AI-based interactive reading material personalized recommendation method and system |
| CN113935332A (en) * | 2021-11-02 | 2022-01-14 | 海信集团控股股份有限公司 | Book grading method and book grading equipment |
| CN114818921A (en) * | 2022-04-26 | 2022-07-29 | 北京思源智通科技有限责任公司 | Multi-dimensional book grading method, system, readable medium and equipment |
| CN115630140A (en) * | 2022-11-02 | 2023-01-20 | 云南师范大学 | English reading material difficulty judgment method based on text feature fusion |
| CN116401466A (en) * | 2023-06-08 | 2023-07-07 | 北京奇趣万物科技有限公司 | Book classification recommendation method and system |
Non-Patent Citations (1)
| Title |
|---|
| 基于图书语义特征的推荐模型;刘园园和李雅琴;《武汉工程大学学报》;第45卷(第03期);第321-322页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116796045A (en) | 2023-09-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116796045B (en) | Multi-dimensional book grading method, system and readable medium | |
| CN111325571B (en) | Automatic generation method, device and system for commodity comment labels for multitask learning | |
| CN110874439B (en) | Recommendation method based on comment information | |
| CN110751188B (en) | User label prediction method, system and storage medium based on multi-label learning | |
| CN111708882A (en) | Transformer-based Chinese text information missing completion method | |
| CN115390806A (en) | Software design mode recommendation method based on bimodal joint modeling | |
| CN116911929A (en) | Advertisement service terminal and method based on big data | |
| CN116663566A (en) | Aspect-level emotion analysis method and system based on commodity evaluation | |
| CN117473034A (en) | Interactive text processing method and device, electronic equipment and storage medium | |
| CN113741759B (en) | Comment information display method and device, computer equipment and storage medium | |
| Chaudhary et al. | Signnet ii: A transformer-based two-way sign language translation model | |
| CN115186085A (en) | Reply content processing method and interaction method of media content interaction content | |
| CN116958997B (en) | Graphic summary method and system based on heterogeneous graphic neural network | |
| Ahmed et al. | Context based emotion recognition from bengali text using transformers | |
| CN116956289B (en) | Method for dynamically adjusting potential blacklist and blacklist | |
| CN113486143A (en) | User portrait generation method based on multi-level text representation and model fusion | |
| CN117494760A (en) | Semantic tag-rich data augmentation method based on ultra-large-scale language model | |
| Shah et al. | A study of various word embeddings in deep learning | |
| CN113590745B (en) | Interpretable text inference method | |
| CN114003708A (en) | Automatic question answering method and device based on artificial intelligence, storage medium and server | |
| CN113822047A (en) | Text enhancement method and device, electronic equipment and storage medium | |
| CN113297385A (en) | Multi-label text classification model and classification method based on improved GraphRNN | |
| Yang et al. | Topic-Oriented Controlled Text Generation for Social Networks | |
| Che et al. | A feature and deep learning model recommendation system for mobile application | |
| Sai et al. | Extraction of Emojis and Texts to Intensify Opinion Mining using Machine Learning and Deep Learning 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 |