CN112464101A - Electronic book sorting recommendation method, electronic device and storage medium - Google Patents
Electronic book sorting recommendation method, electronic device and storage medium Download PDFInfo
- Publication number
- CN112464101A CN112464101A CN202011481357.8A CN202011481357A CN112464101A CN 112464101 A CN112464101 A CN 112464101A CN 202011481357 A CN202011481357 A CN 202011481357A CN 112464101 A CN112464101 A CN 112464101A
- Authority
- CN
- China
- Prior art keywords
- interaction
- book
- interactive
- transfer
- weight
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 67
- 230000002452 interceptive effect Effects 0.000 claims abstract description 336
- 230000003993 interaction Effects 0.000 claims abstract description 325
- 238000012549 training Methods 0.000 claims abstract description 138
- 238000012163 sequencing technique Methods 0.000 claims abstract description 63
- 238000013507 mapping Methods 0.000 claims abstract description 60
- 238000012546 transfer Methods 0.000 claims description 303
- 238000004891 communication Methods 0.000 claims description 20
- 238000013016 damping Methods 0.000 claims description 7
- 238000010586 diagram Methods 0.000 claims description 6
- 238000002372 labelling Methods 0.000 claims description 5
- 230000007704 transition Effects 0.000 claims description 4
- 230000006399 behavior Effects 0.000 description 9
- 238000012544 monitoring process Methods 0.000 description 3
- 238000003062 neural network model Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000009191 jumping Effects 0.000 description 2
- 230000001960 triggered effect Effects 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011022 operating instruction Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9538—Presentation of query results
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a sequencing recommendation method of an electronic book, electronic equipment and a storage medium, wherein the method comprises the following steps: acquiring an interactive book list corresponding to an interactive time period of a reading user; determining an interaction depth value of an interaction book according to interaction duration and/or interaction type of the interaction book corresponding to the reading user aiming at the interaction book in the interaction book list; mapping the interactive depth value of the interactive book to an interactive weight value corresponding to the bucket dividing interval according to a preset bucket dividing mapping relation; a plurality of training sample sets are constructed according to the interaction weight values of all interaction books in the interaction book list, a sequencing push model is trained through the training sample sets, and sequencing recommendation of the electronic books is carried out on the basis of the sequencing push model. The method considers the influence of the interaction depth, so that the sequencing result is more accurate. In addition, the interaction depth values which are inconvenient to compare can be converted into the interaction weight values which are convenient to compare in a bucket mapping mode, and therefore a plurality of training sample groups can be conveniently constructed.
Description
Technical Field
The invention relates to the field of computers, in particular to a method for recommending sequencing of electronic books, electronic equipment and a storage medium.
Background
Books in the form of electronic books are popular with a large number of users because of their advantages such as easy access. Most book reading platforms recommend books according to the content similarity of the whole book. In the prior art, a plurality of books having a high similarity to books that a user has read are generally used as recommended books and presented to the user.
However, the inventor finds that the recommended mode has at least the following defects in the process of implementing the invention: recommendation according to the similarity of the book contents easily causes a user to read a large number of books with the same contents, and is not beneficial to expanding the reading range of the user. In addition, in the book recommendation process, targeted recommendation cannot be performed according to the difference of the interaction depth of the user on different books. In a word, the existing book recommendation mode is single, so that the matching degree of the recommended content and the user requirement is not high.
Disclosure of Invention
In view of the above problems, the present invention has been made to provide a sort recommendation method for an electronic book, an electronic device, and a storage medium that overcome or at least partially solve the above problems.
According to an aspect of the present invention, there is provided a method for recommending ordering of an electronic book and pushing ordering of the book, including:
acquiring an interactive book list corresponding to an interactive time period of a reading user;
determining an interaction depth value of the interaction book according to an interaction duration and/or an interaction type of a reading user corresponding to the interaction book aiming at the interaction book in the interaction book list;
mapping the interactive depth value of the interactive book to an interactive weight value corresponding to a bucket dividing interval according to a preset bucket dividing mapping relation;
constructing a plurality of training sample groups according to the interaction weight values of all the interaction books in the interaction book list, training a sequencing push model through the training sample groups, and performing sequencing recommendation on the electronic books based on the sequencing push model.
According to another aspect of the present invention, there is provided an electronic apparatus including: the system comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete mutual communication through the communication bus;
the memory is configured to store at least one executable instruction that causes the processor to:
acquiring an interactive book list corresponding to an interactive time period of a reading user;
determining an interaction depth value of the interaction book according to an interaction duration and/or an interaction type of a reading user corresponding to the interaction book aiming at the interaction book in the interaction book list;
mapping the interactive depth value of the interactive book to an interactive weight value corresponding to a bucket dividing interval according to a preset bucket dividing mapping relation;
constructing a plurality of training sample groups according to the interaction weight values of all the interaction books in the interaction book list, training a sequencing push model through the training sample groups, and performing sequencing recommendation on the electronic books based on the sequencing push model.
According to yet another aspect of the present invention, there is provided a computer storage medium having at least one executable instruction stored therein, the executable instruction causing the processor to:
acquiring an interactive book list corresponding to an interactive time period of a reading user;
determining an interaction depth value of the interaction book according to an interaction duration and/or an interaction type of a reading user corresponding to the interaction book aiming at the interaction book in the interaction book list;
mapping the interactive depth value of the interactive book to an interactive weight value corresponding to a bucket dividing interval according to a preset bucket dividing mapping relation;
constructing a plurality of training sample groups according to the interaction weight values of all the interaction books in the interaction book list, training a sequencing push model through the training sample groups, and performing sequencing recommendation on the electronic books based on the sequencing push model.
In the method for recommending the sequencing of the electronic books, the electronic equipment and the storage medium provided by the invention, firstly, an interactive book list corresponding to an interactive time period of a reading user is obtained, and the interactive depth value of the interactive book is determined aiming at the interactive book in the interactive book list; then, mapping the interactive depth value of the interactive book to an interactive weight value corresponding to the bucket-dividing interval according to a preset bucket-dividing mapping relation; and finally, constructing a training sample set according to the interaction weight values of the interaction books so as to train a sequencing push model, thereby realizing sequencing recommendation of the electronic books. Therefore, the training sample set can be constructed according to the interactive depth values of the interactive books, so that different sequencing recommendations can be carried out on the sequencing push model obtained through training according to different interactive depth values of the books, the influence of the interactive depth is considered in the sequencing recommendation process, and the sequencing result is more accurate. In addition, the interaction depth values which are inconvenient to compare can be converted into the interaction weight values which are convenient to compare in a bucket mapping mode, and therefore a plurality of training sample groups can be conveniently constructed.
The foregoing description is only an overview of the technical solutions of the present invention, and the embodiments of the present invention are described below in order to make the technical means of the present invention more clearly understood and to make the above and other objects, features, and advantages of the present invention more clearly understandable.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
fig. 1 is a flowchart illustrating a method for ordering recommendation of an electronic book and pushing book ordering according to an embodiment of the present invention;
fig. 2 is a flowchart illustrating an ordering recommendation method for an electronic book and a book ordering push method according to another embodiment of the present invention;
fig. 3 shows a schematic structural diagram of an electronic device according to another embodiment of the invention.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
Example one
Fig. 1 shows a flowchart of a method for ordering recommendation of an electronic book and pushing book ordering according to an embodiment of the present invention. As shown in fig. 1, the method comprises the steps of:
step S110: and acquiring an interactive book list corresponding to an interactive period of the reading user.
The interaction period refers to a period of time defined by a starting time period and an ending time period, the user behavior is continuously monitored in the interaction period, and the interaction book list corresponding to the interaction period is obtained according to the monitoring result. The interactive book list is used for storing at least one electronic book interacted by the reading user in the interaction time period. Wherein, generating the interaction comprises: the user triggers various preset types of interactive operations such as clicking operation, searching operation, browsing operation and the like aiming at the electronic book, and the type and the number of the interactive operations can be flexibly set by technicians in the field according to service scenes.
Step S120: and determining the interaction depth value of the interaction book according to the interaction duration and/or the interaction type of the reading user corresponding to the interaction book aiming at the interaction book in the interaction book list.
Specifically, for each interactive book, the interaction duration and/or the interaction type of the reading user corresponding to the interactive book are determined. The interaction duration mainly refers to reading duration, and the longer the reading duration is, the larger the interaction depth value is. In addition, the interaction types include: different weights can be set in advance for different interaction types, and the interaction depth value of the interaction book is determined according to the weighting result of the interaction operation corresponding to each interaction type.
In summary, the interaction depth value of the interaction book is used to reflect the interaction degree of the reading user with respect to the interaction book: the larger the interaction depth value is, the deeper the interaction of the reading user on the interaction book is, namely: the greater the preference, the more interesting.
Step S130: and mapping the interactive depth value of the interactive book to an interactive weight value corresponding to the bucket-dividing interval according to a preset bucket-dividing mapping relation.
Specifically, a bucket mapping relationship is pre-established, and the bucket mapping relationship is used for mapping continuous interaction depth values which are inconvenient to compare into discrete interaction weight values which are convenient to compare. The sub-bucket mapping relation sets a plurality of sub-bucket intervals, and the numerical range of the interactive depth value corresponding to each sub-bucket interval and the interactive weight value corresponding to the sub-bucket interval.
Correspondingly, based on the bucket mapping relationship, the interaction depth value of the interaction book is converted into an interaction weight value corresponding to the bucket interval, so that comparison is performed on the interaction weight value in the subsequent steps.
Step S140: a plurality of training sample sets are constructed according to the interaction weight values of all interaction books in the interaction book list, a sequencing push model is trained through the training sample sets, and sequencing recommendation of the electronic books is carried out on the basis of the sequencing push model.
One training sample group may be a training sample pair including two interactive books, or may be a training sample list including a plurality of interactive books, which is not limited in this disclosure. In specific implementation, the training sample set is labeled according to a comparison result between interaction weight values of a plurality of training samples (i.e., the interaction books) in the training sample set. And training a sequencing pushing model based on the labeled training sample groups. The sequencing pushing model can be various models such as a neural network model, and the invention is not limited to this. When the electronic book is recommended through the sequencing push model, the difference of the interaction depth of the user on different electronic books can be fully considered, and therefore the accuracy of a recommendation result is improved.
Therefore, the training sample set can be constructed according to the interactive depth values of the interactive books, so that different sequencing recommendations can be carried out on the sequencing push model obtained through training according to different interactive depth values of the books, the influence of the interactive depth is considered in the sequencing recommendation process, and the sequencing result is more accurate. In addition, the interaction depth values which are inconvenient to compare can be converted into the interaction weight values which are convenient to compare in a bucket mapping mode, and therefore a plurality of training sample groups can be conveniently constructed.
Example two
Fig. 2 is a flowchart illustrating a method for recommending an electronic book by sorting according to another embodiment of the present invention. As shown in fig. 2, the method comprises the steps of:
step S210: and acquiring an interactive book list corresponding to an interactive period of the reading user.
The interaction period refers to a period of time defined by a starting time period and an ending time period, the user behavior is continuously monitored in the interaction period, and the interaction book list corresponding to the interaction period is obtained according to the monitoring result. The interactive book list is used for storing at least one electronic book interacted by the reading user in the interaction time period. Wherein, generating the interaction comprises: the user triggers various preset types of interactive operations such as clicking operation, searching operation, browsing operation and the like aiming at the electronic book, and the type and the number of the interactive operations can be flexibly set by technicians in the field according to service scenes.
Step S220: and determining the interaction depth value of the interaction book according to the interaction duration and/or the interaction type of the reading user corresponding to the interaction book aiming at the interaction book in the interaction book list.
Specifically, for each interactive book, the interaction duration and/or the interaction type of the reading user corresponding to the interactive book are determined. The interaction duration mainly refers to reading duration, and the longer the reading duration is, the larger the interaction depth value is. In addition, the interaction types include: different weights can be set in advance for different interaction types, and the interaction depth value of the interaction book is determined according to the weighting result of the interaction operation corresponding to each interaction type. In summary, the interaction depth value of the interaction book is used to reflect the interaction degree of the reading user with respect to the interaction book: the larger the interaction depth value is, the deeper the interaction of the reading user on the interaction book is, namely: the greater the preference, the more interesting.
In specific implementation, in order to ensure the comprehensiveness of the sample, the interactive book list needs to be continuously acquired for a plurality of reading users in a plurality of interactive periods. Therefore, in order to distinguish different reading users and different interaction periods, the list identifier of the interactive book list needs to include a user identifier, which may be a user ID of the reading user, and period information, which is identified by the start time point and the end time point. Therefore, each reading user as a sample at least contains one interactive book in the interactive book list in an interactive period. Then, for each interactive book contained in the interactive book list, the interactive duration and/or the interactive type generated by the user for the interactive book within the preset duration are monitored. The preset time duration can be three days or seven days, and since the reading time of a book is long, the preset time duration needs to be set to be slightly longer, and the interaction depth value corresponding to the interaction book by the user is determined by continuously monitoring the interaction time duration and the interaction type of each interaction behavior triggered by the user in the preset time duration for the interaction book. For example, the longer the interaction duration, the larger the interaction depth value; for another example, different weights may be set in advance for each type of interaction behavior, so that a weighting operation is performed on each interaction behavior according to an interaction type, and an interaction depth value is determined according to a weighting result, for example, the weights of the following interaction types decrease sequentially: a paid type interaction type, a download type, a bookcase adding type, a free chapter browsing type, and the like.
Step S230: and mapping the interactive depth value of the interactive book to an interactive weight value corresponding to the bucket-dividing interval according to a preset bucket-dividing mapping relation.
Specifically, a bucket mapping relationship is pre-established, and the bucket mapping relationship is used for mapping continuous interaction depth values which are inconvenient to compare into discrete interaction weight values which are convenient to compare. The sub-bucket mapping relation sets a plurality of sub-bucket intervals, and the numerical range of the interactive depth value corresponding to each sub-bucket interval and the interactive weight value corresponding to the sub-bucket interval. Correspondingly, based on the bucket mapping relationship, the interaction depth value of the interaction book is converted into an interaction weight value corresponding to the bucket interval, so that comparison is performed on the interaction weight value in the subsequent steps.
In specific implementation, the interaction weight value of the interaction book is determined in the following way: firstly, inquiring a preset barrel mapping relation, and determining a barrel interval corresponding to an interactive depth value of an interactive book. The sub-bucket mapping relation is used for setting a numerical range of the interactive depth value corresponding to each sub-bucket interval and an interactive weight value corresponding to each sub-bucket interval. And then, inquiring an interaction weight value corresponding to the sub-bucket interval corresponding to the interaction depth value of the interaction book, and taking the inquired interaction weight value as the interaction weight value of the interaction book. The interactive depth value is a continuous numerical value, and the interactive weight value is a discrete numerical value.
For example, if the reading user does not generate an interaction with respect to the book, the interaction depth value of the interaction book is 0 (specifically marked as a negative sample). When the reading user generates an interaction with respect to the book, the interaction depth value is a positive value (specifically marked as a positive sample), and the specific numerical value is determined according to the interaction duration and/or the interaction type. In specific implementation, 18 data sub-buckets may be set, each data sub-bucket corresponding to a different sub-bucket interval: the bucket interval of the first data bucket is (0-10), the interactive book whose interactive depth value belongs to the interval of (0-10) will fall into the first data bucket, and the interactive weight value corresponding to the first data bucket is 1, and the interactive weight value of the interactive book falling into the first data bucket is 1, the bucket interval of the second data bucket is (10-300), the interactive book whose interactive depth value belongs to the interval of (10-300) will fall into the second data bucket, and the interactive weight value corresponding to the second data bucket is 2, and the interactive book falling into the second data bucket has an interactive weight value of 2 … …, and the interactive weight value of the 18 th data bucket is 18, and the interactive book falling into the 18 th data bucket has an interactive weight value of 18, thus it can be seen that, through the data bucket dividing mode, continuous interaction depth values which are inconvenient to compare are converted into discrete interaction weight values which are convenient to compare.
By means of barrel mapping, the difference between the interaction depths can be better reflected. In the process of implementing the present invention, the inventor finds that, because the interaction depth value is usually determined according to the interaction duration, if the continuous interaction depth values are directly compared, differences between different books cannot be effectively mined. For example, assume that a book interacts for 299 minutes, the corresponding interaction depth value is 299; book two is interacted for 300 minutes, the corresponding interaction depth value is 300, if a training sample group is directly constructed according to the interaction depth value, the difference value of the interaction depth values between book two and book one is 1, but actually, the interaction depths of book one and book two belong to a barrel interval, and the significance of comparing the two is not great.
Step S240: and constructing a plurality of training sample groups according to the interaction weight values of all the interaction books in the interaction book list.
Specifically, each interactive book in the interactive book list is combined into a plurality of training sample groups according to the interactive weight value of each interactive book in the interactive book list, and each training sample group is labeled according to a comparison result between the interactive weight values of at least two training samples included in each training sample group. In specific implementation, a plurality of interactive books with different interactive weight values are determined as a training sample group, and the training sample group is labeled according to a comparison result between the interactive weight values of the plurality of interactive books in the training sample group.
The training sample set in this embodiment may be constructed in various ways, for example, the training sample set may be a training sample pair composed of two interactive book samples, or a training sample list composed of a plurality of interactive book samples, which is not limited in this disclosure.
Taking the training sample pairs as an example, when constructing the training sample pairs, the plurality of positive samples may be sorted according to the interaction weight values, and then each pair of positive samples obtained after sorting is combined to obtain the plurality of training sample pairs. Specifically, because the Pointwise algorithm is more suitable for processing samples with short interaction time, and the fitting performance of samples with long interaction time is not good, in this embodiment, the ranking recommendation model may be trained through the Pairwise algorithm and/or the Listwise algorithm, in order to better adapt to the above algorithm, a plurality of positive samples with interaction depth values greater than a preset depth threshold may be screened, and a training sample pair may be constructed according to the screened positive samples, so as to intensively learn the characteristics of samples with long interaction time. The difference in the interaction depth between samples can be expressed by the form of training sample pairs.
The inventor finds that if the interactive depth values of two books are directly compared, the comparison result cannot truly reflect the difference degree between the two books. For example, assuming that book in the first training sample set interacted for 290 minutes, the corresponding interaction depth value is 290; the book two is interacted for 300 minutes, the corresponding interaction depth value is 300, and if the first training sample group is directly constructed according to the interaction depth value, the difference value of the interaction depth values between the book two and the book one is 10; assuming that books in the second training sample group are interacted for 1 minute, and the corresponding interaction depth value is 1; and if the book four is interacted for 11 minutes, the corresponding interaction depth value is 11, and if the second training sample group is directly constructed according to the interaction depth value, the difference value of the interaction depth values between the book three and the book four is also 10. In the above manner, the difference between the first training sample set and the second training sample set is the same, but in the case of a long interaction time, the difference between the interaction depth of 290 minutes and the interaction depth of 300 minutes is not large; and in the case of a short interaction time, the interaction depth of 1 minute is more different from the interaction depth of 11 minutes. Therefore, when the training sample set is directly constructed by the interactive depth values, the degree of difference between the samples cannot be accurately reflected.
In order to solve the above problem, in this embodiment, a bucket mapping manner is used to map a continuous interaction depth value into a discrete interaction weight value, and after the bucket mapping process, the first book and the second book belong to the same bucket interval together, so that the interaction weight values of the first book and the second book are equal to each other, and at this time, the first book and the second book do not need to be constructed into a training sample group. Therefore, after barrel mapping processing, the constructed training sample set can be ensured to reflect the difference among samples.
Step S250: training a sequencing pushing model through a plurality of training sample sets, and carrying out sequencing recommendation on the electronic book based on the sequencing pushing model.
Specifically, the training sample set is labeled according to a comparison result between interaction weight values of a plurality of training samples (i.e., the interaction books) in the training sample set. And training a sequencing pushing model based on the labeled training sample groups. In specific implementation, a plurality of training sample sets are trained through a ranking algorithm to obtain a ranking pushing model. Wherein, the sequencing algorithm comprises: a sample pair based ordering algorithm and/or a sample list based ordering algorithm.
The ranking push model in this embodiment may be a neural network model, and the neural network model includes a network layer and a full connection layer. Accordingly, the above-mentioned ranking algorithm is mainly used for training at the fully-connected layer.
In addition, the training sample group in this embodiment is mainly used for learning the ranking information between the samples based on the Pairwise algorithm and/or the Listwise algorithm, but in practical application, when the training sample pairs are sparse, the Pairwise algorithm and/or the Listwise algorithm are directly adopted, which may cause the problem of overfitting. In addition, through the pre-training process, the sample vectors and the sample weights corresponding to the training samples obtained in the network layer are obtained, and accordingly, in the fine tuning process, the sample vectors and the sample weights obtained in the pre-training process can be directly shared.
In addition, the inventor finds that the transfer relationship between books can effectively reflect the importance of the books in the process of implementing the invention, so that in an improved implementation manner, in the process of training the sequencing recommendation model, the transfer weight determined according to the transfer relationship between the interactive books is further used as a sample attribute feature of the interactive books, and the accuracy of the model is further improved.
Correspondingly, when a plurality of training sample sets are constructed according to the interaction weight values of the interaction books in the interaction book list, the method is further realized in the following mode: firstly, determining the transfer relationship among all interactive books in an interactive book list, and generating a book transfer graph corresponding to the interactive book list according to the transfer relationship among the interactive books; then, according to the book transfer graph, the transfer weight of each interactive book is calculated, and the transfer weight is used as the sample characteristic of the training sample corresponding to the interactive book. The book transition diagram is generated according to the reading behavior data of the user. The book transfer graph comprises a plurality of transfer nodes, and each transfer node corresponds to one interactive book. Because a user has a behavior of jumping from a first interactive book to a second interactive book in a reading process, a transfer relationship between transfer nodes is established based on a book jumping behavior (also called book transfer behavior) triggered by the user, wherein the transfer relationship includes: the direction of transfer, for example, into or out of.
Specifically, when the transfer relationship between each interactive book in the interactive book list is determined, and a book transfer graph corresponding to the interactive book list is generated according to the transfer relationship between the interactive books, the book transfer graph is generated by the following method: detecting whether a reading user interacts with a second interactive book or not in the process of reading the first interactive book aiming at the first interactive book in the interactive book list; if so, determining a second transfer node corresponding to the second interactive book as a transfer-out type transfer node of a first transfer node corresponding to the first interactive book; and determining a first transfer node corresponding to the first interactive book as a transfer class transfer node of a second transfer node corresponding to the second interactive book. The first interactive book may be any book in the interactive book list, and in a specific implementation, an interactive time period during which a reading user interacts with the first interactive book needs to be determined, where the interactive time period may be in units of hours or days, and the book belongs to a long-interaction business object, so the interactive time period lasts for a long time. Also, during this interaction period, the user may interact with other books incrementally. For example, in the process of reading the western notes, the user downloads the three kingdoms rehearsal or the water entermorphism for alternate reading. Correspondingly, the number of the second interactive books can be one or more, and all newly added interactive books are used as the second interactive books in the interactive time period of reading the first interactive books by the reading user. Therefore, the user turns to the second interactive book from the first interactive book in the reading process, so that the transfer node corresponding to the second interactive book is the transfer-out type transfer node for the first interactive book; for the second interactive book, the transfer node corresponding to the first interactive book is the transfer-in transfer node. For a transfer node corresponding to any interactive book, the transfer node may include both a plurality of transfer-in transfer nodes and a plurality of transfer-out transfer nodes.
Correspondingly, when the transfer weight of each interactive book is calculated according to the book transfer graph, a transfer node corresponding to any interactive book contained in the book transfer graph is used as a target transfer node, and a transfer-in type transfer node corresponding to the target transfer node and a transfer-out type transfer node corresponding to the target transfer node are determined; and calculating the transfer weight of the interactive book corresponding to the target transfer node according to the transfer-in type transfer node and the transfer-out type transfer node. In addition, the number of the transfer-out type transfer nodes of each transfer-in type transfer node and the transfer weight of the transfer-out type transfer node of each transfer-in type transfer node can be further determined respectively aiming at each transfer-in type transfer node corresponding to the target transfer node, and the transfer weight of the interactive book corresponding to the target transfer node can be calculated by combining the number of the transfer-out type transfer nodes of each transfer-in type transfer node and the transfer weight of the transfer-out type transfer node of each transfer-in type transfer node.
In specific implementation, the transfer weight of the interactive book corresponding to the target transfer node may be calculated according to a preset damping coefficient, the total number of transfer nodes included in the book transfer graph, the transfer weight of each transfer-in type transfer node corresponding to the target transfer node, and/or the transfer weight of each transfer-out type transfer node corresponding to the target transfer node. For example, the transfer weight of the interactive book corresponding to the target transfer node may be calculated according to a ratio between the sum of the transfer weights of the transfer-in transfer nodes corresponding to the target transfer node and the sum of the transfer weights of the transfer-out transfer nodes corresponding to the target transfer node. Of course, those skilled in the art can also flexibly set various ways to calculate the transfer weight of the interactive book corresponding to the target transfer node, and the specific calculation details are not limited in the present invention.
In an alternative implementation manner, the transfer weight pr (u) of the interactive book corresponding to the target transfer node u may be calculated by the following formula:
wherein d is a damping coefficient, and the specific numerical value can be flexibly set according to the actual condition. And N is the total number of the transfer nodes contained in the book transfer graph. In practical cases, N may be the total number of all the transfer nodes included in the book transfer graph, or may be the number of associated transfer nodes having a transfer-in or output relationship with the target transfer node. In the latter mode, the value of N is determined according to the number of associated transfer nodes corresponding to the target transfer node. Wherein associating the transfer node comprises: direct association transfer nodes and indirect association transfer nodes. Specifically, directly associating the transfer node means: a transfer-in class transfer node and a transfer-out class transfer node corresponding to the target transfer node; the indirect association of the transfer node means: a transfer-in class transfer node and/or a transfer-out class transfer node corresponding to a directly associated transfer node of the target transfer node.
Wherein V in the above formula represents any transfer node in the set Iu, the set Iu is used to store each transfer-in type transfer node corresponding to the target transfer node, pr (V) represents the transfer weight of the transfer node V, and | o (V) | represents the number of transfer-out type transfer nodes corresponding to the transfer node V. Through the formula, the transfer weight value of the interactive book corresponding to the target transfer node u can be calculated.
It should be further emphasized that the above-mentioned interactive book list may be a list corresponding to one reading user, or may be a list corresponding to a plurality of reading users, which is not limited in the present invention. Correspondingly, when the interactive book list is a list corresponding to a plurality of reading users, the book transfer diagram can reflect the transfer conditions of the users, so that the result is more in line with the requirements of most users.
In conclusion, the training sample set can be constructed according to the interaction depth values of the interaction books, so that different sequencing recommendations can be carried out on the sequencing push model obtained through training according to different interaction depth values of the books, the influence of the interaction depth is considered in the sequencing recommendation process, and the sequencing result is more accurate. In addition, the interaction depth values which are inconvenient to compare can be converted into the interaction weight values which are convenient to compare in a bucket mapping mode, and therefore a plurality of training sample groups can be conveniently constructed. In addition, in the embodiment, the transfer relationship among the books is further considered in the training process, so that the importance degree of the books is conveniently and accurately determined according to the transfer relationship among the books, and the sequencing result is more accurate.
EXAMPLE III
The embodiment of the application provides a nonvolatile computer storage medium, wherein the computer storage medium stores at least one executable instruction, and the computer executable instruction can execute the method for recommending the ordering of the electronic book and pushing the ordering of the book in any method embodiment.
The executable instructions may be specifically configured to cause the processor to:
acquiring an interactive book list corresponding to an interactive time period of a reading user;
determining an interaction depth value of the interaction book according to an interaction duration and/or an interaction type of a reading user corresponding to the interaction book aiming at the interaction book in the interaction book list;
mapping the interactive depth value of the interactive book to an interactive weight value corresponding to a bucket dividing interval according to a preset bucket dividing mapping relation;
constructing a plurality of training sample groups according to the interaction weight values of all the interaction books in the interaction book list, training a sequencing push model through the training sample groups, and performing sequencing recommendation on the electronic books based on the sequencing push model.
In an alternative implementation, the executable instructions cause the processor to:
inquiring a preset barrel mapping relation, and determining a barrel interval corresponding to the interactive depth value of the interactive book; the sub-bucket mapping relation is used for setting a numerical range of interactive depth values corresponding to each sub-bucket interval and interactive weight values corresponding to each sub-bucket interval;
inquiring an interaction weight value corresponding to a sub-bucket interval corresponding to the interaction depth value of the interaction book, and taking the inquired interaction weight value as the interaction weight value of the interaction book;
the interaction depth value is a continuous numerical value, and the interaction weight value is a discrete numerical value.
In an alternative implementation, the executable instructions cause the processor to:
and combining each interactive book in the interactive book list into a plurality of training sample groups according to the interactive weight value of each interactive book in the interactive book list, and labeling each training sample group according to a comparison result between the interactive weight values of at least two training samples contained in each training sample group.
In an alternative implementation, the executable instructions cause the processor to:
training the training sample sets through a ranking algorithm to obtain the ranking pushing model; wherein the ranking algorithm comprises: a sample pair based ordering algorithm and/or a sample list based ordering algorithm.
In an alternative implementation, the executable instructions cause the processor to:
determining a transfer relationship between each interactive book in an interactive book list, and generating a book transfer graph corresponding to the interactive book list according to the transfer relationship between the interactive books;
and calculating the transfer weight of each interactive book according to the book transfer graph, and taking the transfer weight as the sample characteristic of the training sample corresponding to the interactive book.
In an alternative implementation, the executable instructions cause the processor to:
detecting whether a reading user interacts with a second interactive book in the process of reading the first interactive book or not aiming at the first interactive book in the interactive book list;
if so, determining a second transfer node corresponding to the second interactive book as a transfer-out type transfer node of a first transfer node corresponding to the first interactive book;
and determining a first transfer node corresponding to the first interactive book as a transfer class transfer node of a second transfer node corresponding to the second interactive book.
In an alternative implementation, the executable instructions cause the processor to:
taking a transfer node corresponding to any interactive book contained in the book transfer graph as a target transfer node, and determining a transfer-in type transfer node corresponding to the target transfer node and a transfer-out type transfer node corresponding to the target transfer node;
and calculating the transfer weight of the interactive book corresponding to the target transfer node according to the transfer-in type transfer node and the transfer-out type transfer node.
In an alternative implementation, the executable instructions cause the processor to:
and calculating the transfer weight of the interactive book corresponding to the target transfer node according to a preset damping coefficient, the total number of transfer nodes contained in the book transfer graph, the transfer weight of each transfer-in type transfer node corresponding to the target transfer node and/or the transfer weight of each transfer-out type transfer node corresponding to the target transfer node.
Example four
Fig. 3 is a schematic structural diagram of an electronic device according to another embodiment of the present invention, and the specific embodiment of the present invention does not limit the specific implementation of the electronic device.
As shown in fig. 3, the electronic device may include: a processor (processor)302, a communication Interface 304, a memory 306, and a communication bus 308.
Wherein: the processor 302, communication interface 304, and memory 306 communicate with each other via a communication bus 308. A communication interface 304 for communicating with network elements of other devices, such as clients or other servers. The processor 302 is configured to execute the program 310, and may specifically execute the ordering recommendation of the electronic book and related steps in the book ordering pushing method embodiment.
In particular, program 310 may include program code comprising computer operating instructions.
The processor 302 may be a central processing unit CPU, or an Application Specific Integrated Circuit (ASIC), or one or more Integrated circuits configured to implement an embodiment of the present invention. The electronic device comprises one or more processors, which can be the same type of processor, such as one or more CPUs; or may be different types of processors such as one or more CPUs and one or more ASICs.
And a memory 306 for storing a program 310. Memory 306 may comprise high-speed RAM memory and may also include non-volatile memory (non-volatile memory), such as at least one disk memory.
The program 310 may specifically be configured to cause the processor 302 to perform the following operations:
acquiring an interactive book list corresponding to an interactive time period of a reading user;
determining an interaction depth value of the interaction book according to an interaction duration and/or an interaction type of a reading user corresponding to the interaction book aiming at the interaction book in the interaction book list;
mapping the interactive depth value of the interactive book to an interactive weight value corresponding to a bucket dividing interval according to a preset bucket dividing mapping relation;
constructing a plurality of training sample groups according to the interaction weight values of all the interaction books in the interaction book list, training a sequencing push model through the training sample groups, and performing sequencing recommendation on the electronic books based on the sequencing push model.
In an alternative implementation, the executable instructions cause the processor to:
inquiring a preset barrel mapping relation, and determining a barrel interval corresponding to the interactive depth value of the interactive book; the sub-bucket mapping relation is used for setting a numerical range of interactive depth values corresponding to each sub-bucket interval and interactive weight values corresponding to each sub-bucket interval;
inquiring an interaction weight value corresponding to a sub-bucket interval corresponding to the interaction depth value of the interaction book, and taking the inquired interaction weight value as the interaction weight value of the interaction book;
the interaction depth value is a continuous numerical value, and the interaction weight value is a discrete numerical value.
In an alternative implementation, the executable instructions cause the processor to:
and combining each interactive book in the interactive book list into a plurality of training sample groups according to the interactive weight value of each interactive book in the interactive book list, and labeling each training sample group according to a comparison result between the interactive weight values of at least two training samples contained in each training sample group.
In an alternative implementation, the executable instructions cause the processor to:
training the training sample sets through a ranking algorithm to obtain the ranking pushing model; wherein the ranking algorithm comprises: a sample pair based ordering algorithm and/or a sample list based ordering algorithm.
In an alternative implementation, the executable instructions cause the processor to:
determining a transfer relationship between each interactive book in an interactive book list, and generating a book transfer graph corresponding to the interactive book list according to the transfer relationship between the interactive books;
and calculating the transfer weight of each interactive book according to the book transfer graph, and taking the transfer weight as the sample characteristic of the training sample corresponding to the interactive book.
In an alternative implementation, the executable instructions cause the processor to:
detecting whether a reading user interacts with a second interactive book in the process of reading the first interactive book or not aiming at the first interactive book in the interactive book list;
if so, determining a second transfer node corresponding to the second interactive book as a transfer-out type transfer node of a first transfer node corresponding to the first interactive book;
and determining a first transfer node corresponding to the first interactive book as a transfer class transfer node of a second transfer node corresponding to the second interactive book.
In an alternative implementation, the executable instructions cause the processor to:
taking a transfer node corresponding to any interactive book contained in the book transfer graph as a target transfer node, and determining a transfer-in type transfer node corresponding to the target transfer node and a transfer-out type transfer node corresponding to the target transfer node;
and calculating the transfer weight of the interactive book corresponding to the target transfer node according to the transfer-in type transfer node and the transfer-out type transfer node.
In an alternative implementation, the executable instructions cause the processor to:
and calculating the transfer weight of the interactive book corresponding to the target transfer node according to a preset damping coefficient, the total number of transfer nodes contained in the book transfer graph, the transfer weight of each transfer-in type transfer node corresponding to the target transfer node and/or the transfer weight of each transfer-out type transfer node corresponding to the target transfer node.
The algorithms and displays presented herein are not inherently related to any particular computer, virtual machine, or other apparatus. Various general purpose systems may also be used with the teachings herein. The required structure for constructing such a system will be apparent from the description above. Moreover, the present invention is not directed to any particular programming language. It is appreciated that a variety of programming languages may be used to implement the teachings of the present invention as described herein, and any descriptions of specific languages are provided above to disclose the best mode of the invention.
In the description provided herein, numerous specific details are set forth. It is understood, however, that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.
Similarly, it should be appreciated that in the foregoing description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of one or more of the various inventive aspects. However, the disclosed method should not be interpreted as reflecting an intention that: that the invention as claimed requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.
Those skilled in the art will appreciate that the modules in the device in an embodiment may be adaptively changed and disposed in one or more devices different from the embodiment. The modules or units or components of the embodiments may be combined into one module or unit or component, and furthermore they may be divided into a plurality of sub-modules or sub-units or sub-components. All of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and all of the processes or elements of any method or apparatus so disclosed, may be combined in any combination, except combinations where at least some of such features and/or processes or elements are mutually exclusive. Each feature disclosed in this specification (including any accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.
Furthermore, those skilled in the art will appreciate that while some embodiments herein include some features included in other embodiments, rather than other features, combinations of features of different embodiments are meant to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments may be used in any combination.
It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The usage of the words first, second and third, etcetera do not indicate any ordering. These words may be interpreted as names.
The invention also discloses A1. an electronic book sequencing recommendation method, which comprises the following steps:
acquiring an interactive book list corresponding to an interactive time period of a reading user;
determining an interaction depth value of the interaction book according to an interaction duration and/or an interaction type of a reading user corresponding to the interaction book aiming at the interaction book in the interaction book list;
mapping the interactive depth value of the interactive book to an interactive weight value corresponding to a bucket dividing interval according to a preset bucket dividing mapping relation;
constructing a plurality of training sample groups according to the interaction weight values of all the interaction books in the interaction book list, training a sequencing push model through the training sample groups, and performing sequencing recommendation on the electronic books based on the sequencing push model.
A2. The method according to a1, wherein the mapping, according to a preset sub-bucket mapping relationship, the interaction depth value of the interaction book to an interaction weight value corresponding to a sub-bucket interval includes:
inquiring a preset barrel mapping relation, and determining a barrel interval corresponding to the interactive depth value of the interactive book; the sub-bucket mapping relation is used for setting a numerical range of interactive depth values corresponding to each sub-bucket interval and interactive weight values corresponding to each sub-bucket interval;
inquiring an interaction weight value corresponding to a sub-bucket interval corresponding to the interaction depth value of the interaction book, and taking the inquired interaction weight value as the interaction weight value of the interaction book;
the interaction depth value is a continuous numerical value, and the interaction weight value is a discrete numerical value.
A3. The method according to a1 or 2, wherein the constructing a plurality of training sample sets according to the interaction weight values of the respective interaction books in the interaction book list comprises:
and combining each interactive book in the interactive book list into a plurality of training sample groups according to the interactive weight value of each interactive book in the interactive book list, and labeling each training sample group according to a comparison result between the interactive weight values of at least two training samples contained in each training sample group.
A4. The method of any of a1-3, wherein the training of the ranked push model over the plurality of training sample sets comprises:
training the training sample sets through a ranking algorithm to obtain the ranking pushing model; wherein the ranking algorithm comprises: a sample pair based ordering algorithm and/or a sample list based ordering algorithm.
A5. The method according to any one of a1-4, wherein the constructing a plurality of training sample sets according to the interaction weight values of the respective interaction books in the interaction book list further includes:
determining a transfer relationship between each interactive book in an interactive book list, and generating a book transfer graph corresponding to the interactive book list according to the transfer relationship between the interactive books;
and calculating the transfer weight of each interactive book according to the book transfer graph, and taking the transfer weight as the sample characteristic of the training sample corresponding to the interactive book.
A6. The method according to a5, wherein the determining a transfer relationship between each interactive book in the interactive book list, and the generating a book transfer diagram corresponding to the interactive book list according to the transfer relationship between the interactive books includes:
detecting whether a reading user interacts with a second interactive book in the process of reading the first interactive book or not aiming at the first interactive book in the interactive book list;
if so, determining a second transfer node corresponding to the second interactive book as a transfer-out type transfer node of a first transfer node corresponding to the first interactive book;
and determining a first transfer node corresponding to the first interactive book as a transfer class transfer node of a second transfer node corresponding to the second interactive book.
A7. The method according to a6, wherein the calculating the transfer weight of each interactive book according to the book transfer graph comprises:
taking a transfer node corresponding to any interactive book contained in the book transfer graph as a target transfer node, and determining a transfer-in type transfer node corresponding to the target transfer node and a transfer-out type transfer node corresponding to the target transfer node;
and calculating the transfer weight of the interactive book corresponding to the target transfer node according to the transfer-in type transfer node and the transfer-out type transfer node.
A8. The method according to a7, wherein the calculating a transfer weight of the interactive book corresponding to the target transfer node according to the transfer-in transfer node and the transfer-out transfer node includes:
and calculating the transfer weight of the interactive book corresponding to the target transfer node according to a preset damping coefficient, the total number of transfer nodes contained in the book transfer graph, the transfer weight of each transfer-in type transfer node corresponding to the target transfer node and/or the transfer weight of each transfer-out type transfer node corresponding to the target transfer node.
B9. An electronic device, comprising: the system comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete mutual communication through the communication bus;
the memory is configured to store at least one executable instruction that causes the processor to:
acquiring an interactive book list corresponding to an interactive time period of a reading user;
determining an interaction depth value of the interaction book according to an interaction duration and/or an interaction type of a reading user corresponding to the interaction book aiming at the interaction book in the interaction book list;
mapping the interactive depth value of the interactive book to an interactive weight value corresponding to a bucket dividing interval according to a preset bucket dividing mapping relation;
constructing a plurality of training sample groups according to the interaction weight values of all the interaction books in the interaction book list, training a sequencing push model through the training sample groups, and performing sequencing recommendation on the electronic books based on the sequencing push model.
B10. The electronic device of B9, wherein the executable instructions cause the processor to:
inquiring a preset barrel mapping relation, and determining a barrel interval corresponding to the interactive depth value of the interactive book; the sub-bucket mapping relation is used for setting a numerical range of interactive depth values corresponding to each sub-bucket interval and interactive weight values corresponding to each sub-bucket interval;
inquiring an interaction weight value corresponding to a sub-bucket interval corresponding to the interaction depth value of the interaction book, and taking the inquired interaction weight value as the interaction weight value of the interaction book;
the interaction depth value is a continuous numerical value, and the interaction weight value is a discrete numerical value.
B11. The electronic device of B10, wherein the executable instructions cause the processor to:
and combining each interactive book in the interactive book list into a plurality of training sample groups according to the interactive weight value of each interactive book in the interactive book list, and labeling each training sample group according to a comparison result between the interactive weight values of at least two training samples contained in each training sample group.
B12. The electronic device of any of B9-11, wherein the executable instructions cause the processor to:
training the training sample sets through a ranking algorithm to obtain the ranking pushing model; wherein the ranking algorithm comprises: a sample pair based ordering algorithm and/or a sample list based ordering algorithm.
B13. The electronic device of any of B9-12, wherein the executable instructions cause the processor to:
determining a transfer relationship between each interactive book in an interactive book list, and generating a book transfer graph corresponding to the interactive book list according to the transfer relationship between the interactive books;
and calculating the transfer weight of each interactive book according to the book transfer graph, and taking the transfer weight as the sample characteristic of the training sample corresponding to the interactive book.
B14. The electronic device of B13, wherein the executable instructions cause the processor to:
detecting whether a reading user interacts with a second interactive book in the process of reading the first interactive book or not aiming at the first interactive book in the interactive book list;
if so, determining a second transfer node corresponding to the second interactive book as a transfer-out type transfer node of a first transfer node corresponding to the first interactive book;
and determining a first transfer node corresponding to the first interactive book as a transfer class transfer node of a second transfer node corresponding to the second interactive book.
B15. The electronic device of B14, wherein the executable instructions cause the processor to:
taking a transfer node corresponding to any interactive book contained in the book transfer graph as a target transfer node, and determining a transfer-in type transfer node corresponding to the target transfer node and a transfer-out type transfer node corresponding to the target transfer node;
and calculating the transfer weight of the interactive book corresponding to the target transfer node according to the transfer-in type transfer node and the transfer-out type transfer node.
B16. The electronic device of B15, wherein the executable instructions cause the processor to:
and calculating the transfer weight of the interactive book corresponding to the target transfer node according to a preset damping coefficient, the total number of transfer nodes contained in the book transfer graph, the transfer weight of each transfer-in type transfer node corresponding to the target transfer node and/or the transfer weight of each transfer-out type transfer node corresponding to the target transfer node.
C17. A computer storage medium having stored therein at least one executable instruction that causes a processor to perform the method of any one of claims 1-8.
Claims (10)
1. A method for recommending the sequencing of an electronic book comprises the following steps:
acquiring an interactive book list corresponding to an interactive time period of a reading user;
determining an interaction depth value of the interaction book according to an interaction duration and/or an interaction type of a reading user corresponding to the interaction book aiming at the interaction book in the interaction book list;
mapping the interactive depth value of the interactive book to an interactive weight value corresponding to a bucket dividing interval according to a preset bucket dividing mapping relation;
constructing a plurality of training sample groups according to the interaction weight values of all the interaction books in the interaction book list, training a sequencing push model through the training sample groups, and performing sequencing recommendation on the electronic books based on the sequencing push model.
2. The method of claim 1, wherein the mapping the interaction depth value of the interaction book to an interaction weight value corresponding to a bucket interval according to a preset bucket mapping relationship comprises:
inquiring a preset barrel mapping relation, and determining a barrel interval corresponding to the interactive depth value of the interactive book; the sub-bucket mapping relation is used for setting a numerical range of interactive depth values corresponding to each sub-bucket interval and interactive weight values corresponding to each sub-bucket interval;
inquiring an interaction weight value corresponding to a sub-bucket interval corresponding to the interaction depth value of the interaction book, and taking the inquired interaction weight value as the interaction weight value of the interaction book;
the interaction depth value is a continuous numerical value, and the interaction weight value is a discrete numerical value.
3. The method of claim 1 or 2, wherein the constructing a plurality of training sample sets according to the interaction weight values of the respective interaction books in the list of interaction books comprises:
and combining each interactive book in the interactive book list into a plurality of training sample groups according to the interactive weight value of each interactive book in the interactive book list, and labeling each training sample group according to a comparison result between the interactive weight values of at least two training samples contained in each training sample group.
4. The method of any of claims 1-3, wherein the training of the ranked push model over the plurality of training sample sets comprises:
training the training sample sets through a ranking algorithm to obtain the ranking pushing model; wherein the ranking algorithm comprises: a sample pair based ordering algorithm and/or a sample list based ordering algorithm.
5. The method of any one of claims 1-4, wherein the constructing a plurality of training sample sets according to interaction weight values of respective interaction books in the list of interaction books further comprises:
determining a transfer relationship between each interactive book in an interactive book list, and generating a book transfer graph corresponding to the interactive book list according to the transfer relationship between the interactive books;
and calculating the transfer weight of each interactive book according to the book transfer graph, and taking the transfer weight as the sample characteristic of the training sample corresponding to the interactive book.
6. The method of claim 5, wherein the determining a transition relationship between each interactive book in the interactive book list, and the generating a book transition diagram corresponding to the interactive book list according to the transition relationship between the interactive books comprises:
detecting whether a reading user interacts with a second interactive book in the process of reading the first interactive book or not aiming at the first interactive book in the interactive book list;
if so, determining a second transfer node corresponding to the second interactive book as a transfer-out type transfer node of a first transfer node corresponding to the first interactive book;
and determining a first transfer node corresponding to the first interactive book as a transfer class transfer node of a second transfer node corresponding to the second interactive book.
7. The method of claim 6, wherein the calculating a transfer weight for each interactive book according to the book transfer graph comprises:
taking a transfer node corresponding to any interactive book contained in the book transfer graph as a target transfer node, and determining a transfer-in type transfer node corresponding to the target transfer node and a transfer-out type transfer node corresponding to the target transfer node;
and calculating the transfer weight of the interactive book corresponding to the target transfer node according to the transfer-in type transfer node and the transfer-out type transfer node.
8. The method according to claim 7, wherein the calculating a transfer weight of the interactive book corresponding to the target transfer node according to the transfer-in transfer node and the transfer-out transfer node comprises:
and calculating the transfer weight of the interactive book corresponding to the target transfer node according to a preset damping coefficient, the total number of transfer nodes contained in the book transfer graph, the transfer weight of each transfer-in type transfer node corresponding to the target transfer node and/or the transfer weight of each transfer-out type transfer node corresponding to the target transfer node.
9. An electronic device, comprising: the system comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete mutual communication through the communication bus;
the memory is configured to store at least one executable instruction that causes the processor to:
acquiring an interactive book list corresponding to an interactive time period of a reading user;
determining an interaction depth value of the interaction book according to an interaction duration and/or an interaction type of a reading user corresponding to the interaction book aiming at the interaction book in the interaction book list;
mapping the interactive depth value of the interactive book to an interactive weight value corresponding to a bucket dividing interval according to a preset bucket dividing mapping relation;
constructing a plurality of training sample groups according to the interaction weight values of all the interaction books in the interaction book list, training a sequencing push model through the training sample groups, and performing sequencing recommendation on the electronic books based on the sequencing push model.
10. A computer storage medium having stored therein at least one executable instruction that causes a processor to perform the method of any one of claims 1-8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011481357.8A CN112464101A (en) | 2020-12-15 | 2020-12-15 | Electronic book sorting recommendation method, electronic device and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011481357.8A CN112464101A (en) | 2020-12-15 | 2020-12-15 | Electronic book sorting recommendation method, electronic device and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112464101A true CN112464101A (en) | 2021-03-09 |
Family
ID=74802976
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011481357.8A Pending CN112464101A (en) | 2020-12-15 | 2020-12-15 | Electronic book sorting recommendation method, electronic device and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112464101A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113987353A (en) * | 2021-10-29 | 2022-01-28 | 掌阅科技股份有限公司 | Book recommendation method, computing device and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108520076A (en) * | 2018-04-19 | 2018-09-11 | 掌阅科技股份有限公司 | E-book recommends method, electronic equipment and computer storage media |
CN109815406A (en) * | 2019-01-31 | 2019-05-28 | 腾讯科技(深圳)有限公司 | A kind of data processing, information recommendation method and device |
US20190205704A1 (en) * | 2017-12-29 | 2019-07-04 | Guangdong Oppo Mobile Telecommunications Corp., Ltd. | Method for Training Model and Information Recommendation System |
CN111125543A (en) * | 2019-12-17 | 2020-05-08 | 掌阅科技股份有限公司 | Training method of book recommendation sequencing model, computing device and storage medium |
-
2020
- 2020-12-15 CN CN202011481357.8A patent/CN112464101A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20190205704A1 (en) * | 2017-12-29 | 2019-07-04 | Guangdong Oppo Mobile Telecommunications Corp., Ltd. | Method for Training Model and Information Recommendation System |
CN108520076A (en) * | 2018-04-19 | 2018-09-11 | 掌阅科技股份有限公司 | E-book recommends method, electronic equipment and computer storage media |
CN109815406A (en) * | 2019-01-31 | 2019-05-28 | 腾讯科技(深圳)有限公司 | A kind of data processing, information recommendation method and device |
CN111125543A (en) * | 2019-12-17 | 2020-05-08 | 掌阅科技股份有限公司 | Training method of book recommendation sequencing model, computing device and storage medium |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113987353A (en) * | 2021-10-29 | 2022-01-28 | 掌阅科技股份有限公司 | Book recommendation method, computing device and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
TWI689871B (en) | Gradient lifting decision tree (GBDT) model feature interpretation method and device | |
CN107346326B (en) | Method and system for information retrieval | |
US20190164084A1 (en) | Method of and system for generating prediction quality parameter for a prediction model executed in a machine learning algorithm | |
US7689615B2 (en) | Ranking results using multiple nested ranking | |
EP3711000A1 (en) | Regularized neural network architecture search | |
CN111566654A (en) | Machine learning integrating knowledge and natural language processing | |
CN111325338B (en) | Neural network structure evaluation model construction and neural network structure searching method | |
CN109388634B (en) | Address information processing method, terminal device and computer readable storage medium | |
CN111914176B (en) | Question recommendation method and device | |
JP2020512651A (en) | Search method, device, and non-transitory computer-readable storage medium | |
CN110765348B (en) | Hot word recommendation method and device, electronic equipment and storage medium | |
CN113010778A (en) | Knowledge graph recommendation method and system based on user historical interest | |
CN116089567A (en) | Recommendation method, device, equipment and storage medium for search keywords | |
CN112733035A (en) | Knowledge point recommendation method and device based on knowledge graph, storage medium and electronic device | |
CN112380421A (en) | Resume searching method and device, electronic equipment and computer storage medium | |
CN110019751A (en) | Machine learning model modification and natural language processing | |
CN111522886A (en) | Information recommendation method, terminal and storage medium | |
CN113220908A (en) | Knowledge graph matching method and device | |
CN115905687A (en) | Cold start-oriented recommendation system and method based on meta-learning graph neural network | |
CN109635004B (en) | Object description providing method, device and equipment of database | |
CN111160699A (en) | Expert recommendation method and system | |
CN112464101A (en) | Electronic book sorting recommendation method, electronic device and storage medium | |
CN111144098B (en) | Recall method and device for extended question | |
CN112445978A (en) | Electronic book pushing method, electronic equipment and storage medium | |
CN108550019A (en) | A kind of resume selection method and device |
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 |