CN113378063B - Method for determining content diversity based on sliding spectrum decomposition and content sorting method - Google Patents
Method for determining content diversity based on sliding spectrum decomposition and content sorting method Download PDFInfo
- Publication number
- CN113378063B CN113378063B CN202110779586.6A CN202110779586A CN113378063B CN 113378063 B CN113378063 B CN 113378063B CN 202110779586 A CN202110779586 A CN 202110779586A CN 113378063 B CN113378063 B CN 113378063B
- Authority
- CN
- China
- Prior art keywords
- content
- contents
- diversity
- user
- matrix
- 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
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
- G06F16/9537—Spatial or temporal dependent retrieval, e.g. spatiotemporal queries
-
- 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
- G06F16/957—Browsing optimisation, e.g. caching or content distillation
Abstract
The application relates to the technical field of computers and discloses a method for determining content diversity based on sliding spectrum decomposition and a content sorting method, wherein the method for determining content diversity based on sliding spectrum decomposition comprises the following steps: determining T contents; determining a sliding window with a size w according to the display size of the electronic equipment; sliding T contents by using the sliding window based on the time sequence to obtain a content matrix; singular value decomposition is carried out on the content matrix to obtain a plurality of singular values, and the product of the singular values is used as the diversity value of the T contents. The method and the device have the advantages that the user perceives diversification in a long-sequence scene is better captured, the method and the device are more effective in calculation, the time complexity and the space complexity are reduced, and the efficiency is greatly improved.
Description
Technical Field
The present disclosure relates to the field of computer technologies, and in particular, to a method, an apparatus, a device, and a medium for determining content diversity based on sliding spectrum decomposition, and a content sorting method, an apparatus, a device, and a medium.
Background
With the advent of the big data age, the appearance of the personalized recommendation system provides an effective tool for processing the information overload problem, has become the standard of each big platform (e-commerce, information flow, etc.) of the internet, and has made a long-standing development on the technology (personalized recall, personalized ordering, etc.), and gradually transitions from the traditional model to the deep learning age. However, the focus of the current personalized recommendation and related algorithm is mostly to improve the accuracy of the recommendation, but neglect the diversity of the recommendation results, so that a phenomenon that highly similar contents are gathered together easily occurs, namely similar Item is piled up, the interest of the user is limited to a relatively narrow (accurate recommendation with information content of 0) recommendation field, and the user experience is further damaged, especially for users with wide interest and ambiguous demands.
Disclosure of Invention
The embodiment of the application provides a method, a device, equipment and a medium for determining content diversity based on sliding spectrum decomposition.
In a first aspect, an embodiment of the present application provides a method for determining content diversity based on sliding spectrum decomposition, for an electronic device, where the method includes:
Determining T contents;
determining a sliding window with a size w according to the display size of the electronic equipment;
sliding the T contents by the sliding window based on a time sequence to obtain a content matrix, wherein the content matrix comprises T contents expressed by space vectors vi of the contents, i represents the identification of the contents, and the value of i is a positive integer less than or equal to T;
and carrying out singular value decomposition on the content matrix to obtain a plurality of singular values, and taking the product of the singular values as the diversity value of the T contents.
In a possible implementation manner of the first aspect, sliding the T contents with the sliding window based on a time sequence, to obtain a content matrix includes: aggregating w contents in each sliding window into one row to obtain an intermediate matrix A L×w The method comprises the steps of carrying out a first treatment on the surface of the Replacing the elements in the intermediate matrix with the vectors vi to obtain a content matrix X E R L×w×d The method comprises the steps of carrying out a first treatment on the surface of the Wherein l=t-w+m; m represents the number of the contents each time the sliding window slides on the T contents; d represents the dimension of the content vector.
In a possible implementation manner of the first aspect, singular value decomposition is performed on the content matrix X to obtain a plurality of singular values, where the following formula is shown:
Taking the product of a plurality of singular values as the diversity value of the T contents, wherein the product is shown as the following formula:
wherein sigma ijk Representing the singular values; u (u) i (1) ∈R L ，u j (2) ∈R w ，u k (3) ∈R d Respectively representing columns of an orthogonal decomposition matrix of the content matrix X;representing an outer product operation.
In a possible implementation of the first aspect, w is less than or equal to T.
In a possible implementation of the first aspect, the sliding window slides m pieces of content at a time on the T pieces of content, where 1+.m+.w.
In a possible implementation manner of the first aspect, determining a sliding window with a size w according to a display size of the electronic device includes: taking the quantity of the content which can be displayed by the display screen of the electronic equipment as the size w of the sliding window; or based on the number of the content which can be displayed by the display screen of the electronic equipment, expanding the number after a preset amplitude to be used as the size w of the sliding window.
In a second aspect, an embodiment of the present application provides a content sorting method, for an electronic device, where the method includes: screening Z contents according to the user portrait;
n candidate contents related to the user are obtained from the Z contents according to the relevance ranking;
According to any one of the possible methods of the first aspect, determining a diversity of any T contents in the N candidate contents;
and selecting T contents with the largest sum of the relativity and the diversity from the N candidate contents, and sorting.
In a possible implementation of the second aspect, ranking N candidate contents related to the user from the Z contents includes:
and scoring each content of the Z contents according to a preset rule to obtain the correlation between the content and the user, wherein the correlation is shown in the following formula:
wherein T represents a T-th of the T contents;a relevance score representing the T-th of the T content.
In a possible implementation manner of the second aspect, the T contents with the largest sum of the relevance and the diversity in the N candidate contents are selected and ordered, where the T contents are as shown in the following formula:
wherein γ is a hyper-parameter for adjusting the balance of the correlation and the diversity.
In a third aspect, an embodiment of the present application provides an apparatus for determining content diversity based on sliding spectrum decomposition, including:
and a determination module: determining T contents;
determining a sliding window with a size w according to the display size of the electronic equipment;
The processing module is used for: sliding the T contents by using the sliding window based on a time sequence to obtain a content matrix, wherein the content matrix comprises the T contents represented by space vectors vi of the contents, i represents the identification of the contents, and the value of i is a positive integer less than or equal to T;
the calculation module: and carrying out singular value decomposition on the content matrix to obtain a plurality of singular values, and taking the product of the singular values as the diversity value of the T contents.
In a possible implementation manner of the third aspect, sliding the T contents with the sliding window based on a time sequence to obtain a content matrix includes: aggregating w contents in each sliding window into one row to obtain an intermediate matrix A L×w The method comprises the steps of carrying out a first treatment on the surface of the Replacing the elements in the intermediate matrix with the vectors vi to obtain a content matrix X E R L×w×d The method comprises the steps of carrying out a first treatment on the surface of the Wherein l=t-w+m; m represents the number of the contents each time the sliding window slides on the T contents; d represents the dimension of the content vector.
In a possible implementation manner of the third aspect, singular value decomposition is performed on the content matrix X to obtain a plurality of singular values, where the following formula is shown:
Taking the product of a plurality of singular values as the diversity value of the T contents, wherein the product is shown as the following formula:
wherein sigma ijk Representing the singular values; u (u) 1 (1) ∈R L ，u j (2) ∈R w ，u k (3) ∈R d Respectively representing columns of an orthogonal decomposition matrix of the content matrix X;representing an outer product operation.
In a possible implementation of the third aspect, w is less than or equal to T.
In a possible implementation of the third aspect, the sliding window slides m pieces of content at a time on the T pieces of content, where 1.ltoreq.m.ltoreq.w.
In a possible implementation manner of the third aspect, determining a sliding window with a size w according to a display size of the electronic device includes: taking the quantity of the content which can be displayed by the display screen of the electronic equipment as the size w of the sliding window; or based on the number of the content which can be displayed by the display screen of the electronic equipment, expanding the number after a preset amplitude to be used as the size w of the sliding window.
In a fourth aspect, embodiments of the present application provide an apparatus for determining content diversity based on sliding spectrum decomposition, the apparatus for determining content diversity based on sliding spectrum decomposition including:
a memory for storing instructions for execution by one or more processors of the system, an
A processor, one of the processors of the system, for executing the instructions to carry out any one of the possible methods of the first aspect described above.
In a fifth aspect, embodiments of the present application provide a computer readable medium having stored thereon instructions that, when executed on a computer, cause the computer to perform any one of the possible methods of the first aspect described above.
In a sixth aspect, an embodiment of the present application provides a content sorting apparatus, including:
and a screening module: screening Z contents according to the user portrait;
the processing module is used for: n candidate contents related to the user are obtained from the Z contents according to the relevance ranking;
the calculation module: according to any one of the methods of the first aspect, determining diversity of any T content in the N candidate contents;
and a sequencing module: and selecting T contents with the largest sum of the relativity and the diversity from the N candidate contents, and sorting.
In a possible implementation manner of the sixth aspect, ranking N candidate contents related to the user from the Z contents includes:
and scoring each content of the Z contents according to a preset rule to obtain the correlation between the content and the user, wherein the correlation is shown in the following formula:
Wherein T represents a T-th of the T contents;a relevance score representing the T-th of the T content.
In a possible implementation manner of the sixth aspect, the T contents with the largest sum of the relevance and the diversity in the N candidate contents are selected and ordered, where the T contents are as shown in the following formula:
wherein γ is a hyper-parameter for adjusting the balance of the correlation and the diversity.
In a seventh aspect, embodiments of the present application provide a content sorting apparatus, including:
a memory for storing instructions for execution by one or more processors of the system, an
A processor, one of the processors of the system, for executing the instructions to carry out any one of the possible methods of the second aspect described above.
In an eighth aspect, embodiments of the present application provide a computer readable medium having stored thereon instructions that, when executed on a computer, cause the computer to perform any one of the possible methods of the second aspect described above.
According to the technical scheme, the problem of recommendation diversity is studied from the perspective of the content sequence by using a time sequence analysis technology, the content outside the sliding window is combined, and therefore the diversity of the whole project sequence is considered. Compared with the prior art, the technical scheme in the application well captures the diversified perceptions of the user in the long sequence scene, is more effective in calculation, reduces the time complexity and the space complexity, and greatly improves the efficiency. In the online experiment, SSD obtains unexpected technical effects of user browsing duration +0.42%, interaction behavior +0.81%, user browsing richness +0.32% and user experience richness +0.68%.
Drawings
FIG. 1 illustrates a graphical representation of the increase in average content sequence length that a user browses over approximately 1.5 years, according to some embodiments of the present application;
FIG. 2 illustrates a hardware architecture diagram of a method of determining content diversity based on sliding spectrum decomposition, according to some embodiments of the present application;
FIG. 3 illustrates an application scenario diagram of a method of determining content diversity based on sliding spectrum decomposition, according to some embodiments of the present application;
FIG. 4 illustrates a flow chart of a method of determining content diversity based on sliding spectrum decomposition, according to some embodiments of the present application;
FIG. 5 illustrates a schematic diagram of generating a content matrix, according to some embodiments of the present application;
FIG. 6 illustrates a flow chart of a content ordering method, according to some embodiments of the present application;
FIG. 7 illustrates a schematic structural diagram of an apparatus for determining content diversity based on sliding spectrum decomposition, according to some embodiments of the present application;
FIG. 8 illustrates a schematic diagram of a content ordering apparatus, according to some embodiments of the present application;
FIG. 9 illustrates a hardware architecture block diagram of a method of determining content diversity based on sliding spectrum decomposition, according to some embodiments of the present application.
Detailed Description
Illustrative embodiments of the present application include, but are not limited to, a method, apparatus, device, and medium for determining content diversity based on sliding spectrum decomposition and a content ordering method, apparatus, device, and medium.
It is to be appreciated that the methods of determining content diversity based on sliding spectrum decomposition provided herein may be implemented on a variety of electronic devices, including, but not limited to, servers, distributed server clusters of multiple servers, cell phones, tablet computers, laptop computers, desktop computers, wearable devices, head mounted displays, mobile email devices, portable gaming machines, portable music players, reader devices, personal digital assistants, virtual or augmented reality devices, televisions with one or more processors embedded or coupled therein, and the like.
It is to be appreciated that in various embodiments of the present application, the processor may be a microprocessor, a digital signal processor, a microcontroller, or the like, and/or any combination thereof. According to another aspect, the processor may be a single core processor, a multi-core processor, or the like, and/or any combination thereof.
The inventive concepts of the embodiments of the present application are briefly described below.
Currently, two main considerations are made for diversity in personalized recommendation systems: aggregation and individuals. Aggregation considers diversity among all users with the goal of facilitating recommender coverage, i.e., embodying "thousands of people and facets". The individual considers the recommendation diversity of a single user, with the goal of more diversifying the category of recommendation according to the interests of the user. In recent years, in the reordering stage diversity optimization strategy, representative methods in industry are: MRR (Maximal Marginal Relevance), google, youtube and Hulu recommended DPP (Determinantal Point Process), tranformer-based PRM by ali, reinforcement learning-based model SlateQ by Google, youtube, etc., wherein DPP is the most representative, and DPP is used as the reference for comparison experiments in this application, DPP is a probability framework which is introduced for the first time to describe the distribution in heat balance, and is characterized as a determinant of some functions. In the prior art, it is proposed to solve the problem of computational complexity of DPP using a fast greedy algorithm. To improve efficiency in practice, sliding windows play a key role in these methods. In the greedy reasoning process using sliding windows, the content outside the sliding window is ignored. Although the content in the current viewing window has the most direct impact on the user's diverse perception, content outside the sliding window that the user has viewed still has a persistent impact on the user's perception due to the user's memory. The existing diversification method based on the sliding window ignores the content outside the window, so that diversification perception of the user is not completely captured. While expanding the size of the sliding window can solve this problem, it still increases the computation time and actually prevents deployment in production systems with very stringent latency requirements. In view of this, embodiments of the present application provide a method of determining content diversity based on sliding spectrum decomposition, in which: determining T contents; determining a sliding window with a size w according to the display size of the electronic equipment; sliding T contents by using the sliding window based on the time sequence to obtain a content matrix; singular value decomposition is carried out on the content matrix to obtain a plurality of singular values, and the product of the singular values is used as the diversity value of the T contents.
According to the method for determining content diversity based on sliding spectrum decomposition, under the condition that under the view angle of individual users, diversified recommendation is generally converted into a target optimization problem considering correlation and diversity, the problem is fully researched, a scheme for solving the target optimization problem is provided, and a recommendation sequence is modeled as a time sequence observed by the users. By using time series analysis techniques, considering multiple sliding windows to model the diversity of the entire sequence, the best compromise to achieve both recommendation similarity and diversity is given, and this method of determining content diversity based on sliding spectrum decomposition is referred to as SSD (Sliding Spectrum Decomposition). FIG. 1 illustrates a graphical representation of the increase in average content sequence length for a user browsing over the last 1.5 years, as shown in FIG. 1, in which the average length of content sequences recommended by an application viewed by a user according to the disclosed method of determining content diversity based on sliding spectrum decomposition and content ordering method increases by about 50% over the last 1.5 years, i.e., for more content now being recommended by the user tending to view the application, and thus, the information for properly using such content outside of the window can be more consistent with the perception of the user.
After the inventive concept of the embodiments of the present application is introduced, some simple descriptions are made below on application scenarios applicable to the technical solutions of the embodiments of the present application, and it should be noted that the application scenarios described below are only used to illustrate the embodiments of the present application and are not limiting. In the specific implementation process, the technical scheme provided by the embodiment of the application can be flexibly applied according to actual needs.
The technical scheme provided by the embodiment of the application is applicable to common personalized information scenes, such as a recommendation system, advertisement pushing, content retrieval, commodity recommendation and the like, and is mainly exemplified by a note recommendation scene. The note may be a note including at least one of text, picture, video and other multimedia content on a certain platform, and is used for recording the hearts or comments of the user about a certain theme. FIG. 2 illustrates a scene graph for determining content diversity based on sliding spectrum decomposition according to some embodiments of the present application, as shown in FIG. 2, by the method for determining content diversity based on sliding spectrum decomposition and the method for sorting content according to the present application, a user may browse notes with high relevance and diversity, for example, the system may estimate categories of interest of the user including food, travel, reading, pet, etc. by performing the user image for the user analyzed by information such as the user likes food, holds a pet, likes a book, etc. by sliding spectrum decomposition. Performing primary screening based on the note content in the database to obtain 1000 notes, scoring according to the relevance, and intercepting the first 100 notes with the highest scores; and for the 100 notes, calculating the sum of the relevance and the diversity value every 20 notes, selecting 20 notes with the maximum value of the sum as recommended notes recommended to the user, and determining the ordering of the 20 notes based on calculation, for example, the diversity can be displayed in a mode of 'delicious food, travel, reading, pet "," delicious food, travel, reading and pet', and the user is recommended.
The method embodiments provided by the application mode can be executed in hardware involved in implementing the scene, and fig. 3 is a hardware structure block diagram of a method for determining content diversity based on sliding spectrum decomposition according to some embodiments of the application.
The terminal 101 may be a desktop terminal or a mobile terminal, which may be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, portable wearable devices, etc. The terminal 101 may be provided with an application, such as a browser or a user side, that can make a browsing note. The application related to the embodiment of the application can be a software user terminal, a user terminal such as a webpage, an applet and the like, and if the application is a user terminal such as a webpage, an applet and the like, the background server is a background server corresponding to the software, the webpage, the applet and the like, and the specific type of the user terminal is not limited. The user can log in the user on the application to browse the notes, and the method in the embodiment of the application can be utilized to determine the multimedia content recommended to the user while browsing the notes, so that the multimedia content can be displayed on the note interface together or can be displayed on the sliding interface, and the form is not limited to the method. In general, even when a user does not log in, a server corresponding to the user side identifies the user, for example, the user may be identified by a terminal used by the user, and thus the identification may be understood as a user of the user.
The server 102 may be a background server corresponding to an application installed on the terminal 101, for example, may be an independent physical server or a server cluster or a distributed system formed by a plurality of servers, or may be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDNs, basic cloud computing services such as big data and artificial intelligence platforms, but is not limited thereto.
The server 102 can include one or more processors 1021, memory 1022, and I/O interfaces 1023 for interaction with terminals, etc. In addition, the server 102 may be further configured with a database 1024, where the database 1024 may be used to store multidimensional data such as user portraits, note records viewed by the user, text of notes, pictures, labels, and the like. The memory 1022 of the server 102 may further store program instructions of the method for determining content diversity based on sliding spectrum decomposition provided in the embodiment of the present application, where the program instructions can be used to implement the steps of the method for determining content diversity based on sliding spectrum decomposition provided in the embodiment of the present application when executed by the processor 1021 to determine content recommended to a user, and further push the content to a target user, so as to push the content in the terminal 101.
The terminal 101 and the server 102 are connected through a network 103, where the network 103 includes one or more and may include various connection types, such as a wired, wireless communication link, cloud, or optical fiber cable, etc., and the specific examples of the above-mentioned network may include the internet provided by a communication provider of the terminal 101.
First, the processor 1021 reads a note currently viewed by the user stored in the database 1024 corresponding to the terminal 101 through the I/O interface 1023 interacting with the terminal 101, and then the memory 1022 determines the second content by executing the stored program instructions of the method for determining content diversity based on sliding spectrum decomposition and pushes it to the terminal 101 through the I/O interface 1023 interacting with the terminal for presentation to the user.
The following describes in detail a solution for determining content diversity based on sliding spectrum decomposition applied to the hardware shown in fig. 3, according to some embodiments of the present application.
Processor 1021 receives user information from terminal 101 via I/O interface 1023, determines T notes (as examples of T content) from database 1024 for notes of higher relevance to the user that the user estimated from the user representation may be interested in. Next, determining a sliding window with a size w according to the display size of the electronic equipment; sliding T contents by using the sliding window based on the time sequence to obtain a content matrix; singular value decomposition is carried out on the content matrix to obtain a plurality of singular values, and the product of the singular values is used as the diversity value of the T contents. The larger the diversity value, the more categories associated with the user that the T pieces of content relate to, the more comprehensive. According to the technical scheme, the problem of recommendation diversity is studied from the perspective of the content sequence by using a time sequence analysis technology, the content outside the sliding window is combined, and therefore the diversity of the whole project sequence is considered. Compared with the prior art, the technical scheme in the application well captures the diversified perceptions of the user in the long sequence scene, is more effective in calculation, reduces the time complexity and the space complexity, and greatly improves the efficiency.
FIG. 4 illustrates a flow chart of a method of determining content diversity based on sliding spectrum decomposition, according to some embodiments of the present application. As shown in fig. 4, in some embodiments, the method may include:
step 401: t notes are determined.
Specifically, in step 401, in some embodiments, T notes are determined from among the notes that can be provided by the database 1024, specifically, the system determines T notes by the notes that are estimated by the user representation to be of high relevance to the user, specifically, it may be assumed that t=20, that is, it is determined that 20 notes from different categories may be of interest to the user.
Step 402: and determining a sliding window with the size w according to the display size of the electronic equipment.
In particular, first, the size w of the sliding window is generally not larger than T contents, and in particular, it is assumed that w=5, that is, the size of 5 contents is slid over 20 contents at a time. The size of the sliding window in some embodiments depends on the design style of the product. Specifically, for example, a product designs 3 windows, each window has a size of 8, that is, each window displays 8 contents regardless of the size, and the display screen is reduced to reduce the size of each content, but not the display quantity of the content. In addition, the design style of the product can also refer to 'user habit', for example, through investigation of a user, a male can see x pieces of content at a time, and a female can see y pieces of content. Wherein the values of x and y are individually settable. In other words, the design style of the product represents the number of notes that the user can view or perceive. In this application, a display size of an electronic device is taken as an example, and in order to determine a size of a sliding window, the number of notes that can be displayed on a display screen of the electronic device (the content is assumed to have a standard size) is a reference. According to the present application, the size of the sliding window may be directly equal to the number of notes that can be displayed by the display screen of the electronic device. In other embodiments, considering that the amount of content perceived by the user may be a certain amount forward or a certain amount backward, besides the current screen display itself, for example, if the window size is assumed to be 5, the user will not forget the first note when seeing the 6 th note, i.e. the number of the notes perceived by the user is not necessarily related to the display size of the electronic device, so the size w of the sliding window is not limited to the display size of the electronic device, but may also be determined according to the number of notes that the user can browse or perceive.
Step 403: and sliding the T contents by the sliding window based on the time sequence to obtain a content matrix.
Specifically, in step 403, fig. 5 shows a schematic diagram of generating a content matrix according to some embodiments of the present application, as shown in fig. 5, from the perspective of a user, a window slides down along with a user's note, and in the background of the system, as the user slides, the time sequence gradually lengthens; next, 5 notes in each sliding window are aggregated into a row to obtain a two-dimensional intermediate matrix A L×w Wherein the element of matrix A is i 1 ，……i T A reference numeral representing each of the 20 notes; by querying the corresponding table, find the vector representation v corresponding to each note it And replacing the elements in the matrix A with vectors to obtain a content matrix X. It will be appreciated that X is a three-dimensional array of elements v i1 ，……，v iT The spatial representation of each of the 20 notes represents a matrix of all notes that the user would browse. Wherein the table is presented in the form of key_value.
The method has the advantages that the intermediate matrix for representing the label of each note is obtained, and then the label is replaced by the vector representation mode of the corresponding note in space in a table look-up mode, so that the calculated amount is reduced, the system operation efficiency is improved, and the user experience of being free of marks and quick in sensitivity is brought to the client.
Step 404: singular value decomposition is carried out on the content matrix to obtain a plurality of singular values, and the product of the singular values is used as the diversity value of the T contents.
Specifically, in step 404, X is the trace matrix in Singular Spectrum Analysis (SSA) of the univariate time series. In time series analysis, SSA is widely used in various fields such as multivariate statistics, nonlinear dynamics systems, and signal processing. In conventional time series analysis, a complex time series is typically composed of several regular components. For example, grain yield tends to rise with the continued advancement of agriculture, but is also affected by seasons, i.e., the time series of grain yield is the sum of trends and seasonality. SSA is a technique that can decompose a time series into various orthogonal components, where the weights of these components are represented by singular values through singular value decomposition of a trace matrix. In the recommended scenario, we extend the trajectory matrix to the third-order case by embedding the d-dimensional content as a multivariate observation. After SSA, we perform singular value decomposition on the content matrix X according to the following formula:
wherein sigma ijk Representing the singular values; u (u) i (1) ∈R L ，u j (2) ∈R w ，u k (3) ∈R d Respectively representing columns of an orthogonal decomposition matrix of the content matrix X;representing an outer product operation.
Through the above process, the next question is how to define diversity from this decomposition. Let us consider first a simple case when the sliding window slides 5 notes at a time over 20 notes, where m=w, i.e. where there is no overlap between any pair of windows, the windows are independent when diversity is calculated. Therefore we only need to pay attention to a single row of the content matrix X and to retire the track tensor into a matrix. Assume that notes are embedded in an inner product space, i.e., the inner product of a pair of notes can represent relevance. It is understood that the diversity is defined by the volume of the ultra-parallelepiped spanned by these notes, with notes of different analogy spanning a larger volume because of the embedding being more orthogonal. And one of the calculation methods of the matrix volume is to use the cumulative product of the singular values. Based on this we extend this approach to defining the volume of the third order content matrix X as shown in the following formula:
the SSD method in the present application combines a plurality of windows together from the perception of the user when browsing the entire sequence, and the volume of X thus represents the diversity based on the entire sequence and the sliding window, and thus equation (2) is also the product of a plurality of singular values as the diversity value of 20 notes. Thus, we define the diversity of the whole sequence by equations.
In some embodiments, the size w of the sliding window is less than or equal to the determined size of the T notes, specifically, the concept of the sliding window is proposed to capture the perception of the user better and more precisely, and when the size of the sliding window is larger than the size of the notes to be presented to the user, all the T notes can be presented directly at one time, and the existence of the sliding window is not significant.
In some embodiments, the sliding window slides m content at a time over T notes, where 1.ltoreq.m.ltoreq.w. Specifically, assuming that w=5, m=3, i.e. the note of the first sliding window is 12345, and the content of the second sliding window is 45678, so that no note is missed before every two adjacent sliding windows, i.e. it is ensured that 20 notes are slid over at least by one sliding window, so that the content outside the sliding windows is fully and completely considered.
According to the technical scheme, the problem of recommendation diversity is studied from the perspective of the content sequence by using a time sequence analysis technology, the content outside the sliding window is combined, and therefore the diversity of the whole project sequence is considered. Compared with the prior art, the technical scheme in the application well captures the diversified perceptions of the user in the long sequence scene, is more effective in calculation, reduces the time complexity and the space complexity, and greatly improves the efficiency. In the online experiment, SSD obtains unexpected technical effects of user browsing duration +0.42%, interaction behavior +0.81%, user browsing richness +0.32% and user experience richness +0.68%.
According to some embodiments of the present application, a content ordering method 600 is provided, and fig. 6 shows a flow chart of a content ordering method according to some embodiments of the present application. As shown in fig. 6, the method is as follows:
step 601: and screening Z contents according to the user portrait.
Specifically, in step 601, based on the user representation, the processor 1021 performs a preliminary filtering within the system, specifically 1000 notes may be filtered according to the user representation, where the user representation may be derived based on the following information of the user: registration information, avatars, personal data, and the like.
Step 602: n candidate contents relevant to the user are obtained from the Z contents according to the relevance ranking.
Specifically, in step 603, in some embodiments, a machine learning method such as a logistic regression model, a factorization basis, a deep neural network model, etc. is used to sample the notes searched or clicked by the user, and the performance of the user after estimating Z notes to be pushed to the user is fitted to give a score r i Score r i Factors such as the time of the user looking at the note, the click rate, the interaction number (praise, collection, attention to authors and the like), the content of the user's search and the like are comprehensively considered. Specifically, assuming that 100 notes are found out from 1000 notes, each note of the 1000 notes is scored, and the top 100 notes with the highest score are taken as candidate notes. The score does not distinguish between a particular category of notes, but rather a general relevance to the user, in particular the greater the inner product of the vector vi of any two notes, the higher the relevance.
Step 603: and determining the diversity of any T contents in the N candidate contents according to a method for determining the diversity of the contents based on sliding spectrum decomposition.
Specifically, in step 603, the diversity of any T notes of the N candidate notes is determined according to any of the methods of determining content diversity based on sliding spectrum decomposition of the first embodiment. Specifically, 20 out of 100 notes are selected for diversity value calculation, and the method is sharedAnd (5) obtaining results.
Step 604: and selecting T contents with the largest sum of relativity and diversity in the N candidate contents, and sorting.
Specifically, in step 604, in some embodiments, to comprehensively measure the relevance and diversity, the present application proposes to directly sum them, as shown in the following formula:
where γ is a hyper-parameter used to adjust the balance of correlation and diversity. And selecting T pieces of content of the optimal solution comprehensively considering the relevance and the diversity from the N pieces of candidate content, sorting the T pieces of content, and recommending the T pieces of content to the user.
According to the technical scheme, the problem of recommendation diversity is studied from the perspective of the content sequence by using a time sequence analysis technology, the content outside the sliding window is combined, and therefore the diversity of the whole project sequence is considered. Compared with the prior art, the technical scheme in the application well captures the diversified perceptions of the user in the long sequence scene, is more effective in calculation, reduces the time complexity and the space complexity, and greatly improves the efficiency. In the online experiment, SSD obtains unexpected technical effects of user browsing duration +0.42%, interaction behavior +0.81%, user browsing richness +0.32% and user experience richness +0.68%.
According to some embodiments of the present application, there is provided an apparatus 700 for determining content diversity based on a sliding spectrum decomposition, as shown in fig. 7, the apparatus 700 for determining content diversity based on a sliding spectrum decomposition is as follows:
determination module 701: determining T contents; determining a sliding window with a size w according to the display size of the electronic equipment;
the processing module 702: sliding T pieces of content by using a sliding window based on a time sequence to obtain a content matrix, wherein the content matrix comprises T pieces of content represented by space vectors vi of the content, i represents the identification of the content, and the value of i is a positive integer less than or equal to T;
calculation module 703: singular value decomposition is carried out on the content matrix to obtain a plurality of singular values, and the product of the singular values is used as the diversity value of the T contents.
The first embodiment is a method embodiment corresponding to the present embodiment, which may be implemented in cooperation with the first embodiment. The related technical details mentioned in the first embodiment are still valid in this embodiment, and in order to reduce repetition, they are not described here again. Accordingly, the related-art details mentioned in the present embodiment can also be applied to the first embodiment.
A fourth embodiment of the present application relates to an apparatus 801 for determining content diversity based on sliding spectrum decomposition, comprising:
Memory 8011 for storing instructions for execution by one or more processors of the system, an
Processor 8012, one of the processors of the system, is adapted to execute the instructions to implement any one of the possible methods of the first embodiment described above.
The first embodiment is a method embodiment corresponding to the present embodiment, which may be implemented in cooperation with the first embodiment. The related technical details mentioned in the first embodiment are still valid in this embodiment, and in order to reduce repetition, they are not described here again. Accordingly, the related-art details mentioned in the present embodiment can also be applied to the first embodiment.
Specifically, as shown in fig. 8, the device 801 may include one or more (only one is shown in the figure) memories 8011 and a processor 8012 (the processor 8012 may include, but is not limited to, a central processing unit CPU, an image processor GPU, a digital signal processor DSP, a microprocessor MCU, a programmable logic device FPGA, etc.). The specific connection medium between the memory 8011 and the processor 8012 is not limited in the embodiments of the present application. In the embodiment of the present application, the memory 8011 and the processor 8012 are connected by a bus 8013 in fig. 8, the bus 8013 is shown by a bold line in fig. 8, and the connection manner between other components is only schematically illustrated, and is not limited thereto. The bus 8013 may be classified as an address bus, a data bus, a control bus, or the like. For ease of illustration, only one thick line is shown in fig. 8, but not only one bus or one type of bus. It will be appreciated by those skilled in the art that the structure shown in fig. 8 is merely illustrative and is not intended to limit the structure of the electronic device. For example, device 801 may also include more or fewer components than shown in FIG. 8, or have a different configuration than shown in FIG. 8.
The processor 8012 executes various functional applications and data processing by running software programs and modules stored in the memory 8011, i.e., implements the above-described method of determining content diversity based on sliding spectrum decomposition.
The memory 8011 may be used for storing program instructions/modules corresponding to methods of determining content diversity based on sliding spectrum decomposition as in some embodiments of the present application, which are executed by the processor 8012. The memory 8011 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid state memory. In some examples, the memory 8011 may further include memory remotely located relative to the processor 8012, which may be connected to the device via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
A fifth embodiment of the present application relates to a computer storage medium encoded with a computer program, the computer readable medium having stored thereon instructions that, when executed on a computer, cause the computer to perform any one of the possible methods of the first embodiment described above.
The first embodiment is a method embodiment corresponding to the present embodiment, which may be implemented in cooperation with the first embodiment. The related technical details mentioned in the first embodiment are still valid in this embodiment, and in order to reduce repetition, they are not described here again. Accordingly, the related-art details mentioned in the present embodiment can also be applied to the first embodiment.
According to some embodiments of the present application, there is provided a content sorting apparatus 900, as shown in fig. 9, the apparatus 900 is as follows:
screening module 901: screening Z contents according to the user portrait;
processing module 902: n candidate contents relevant to the user are obtained from the Z contents according to the relevance ranking;
calculation module 903: according to any one of the methods of the first aspect, determining a diversity of any T contents of the N candidate contents;
the ranking module 904: and selecting T contents with the largest sum of relativity and diversity in the N candidate contents, and sorting.
The second embodiment is a method embodiment corresponding to the present embodiment, and the present embodiment may be implemented in cooperation with the second embodiment. The related technical details mentioned in the second embodiment are still valid in this embodiment, and in order to reduce repetition, they are not described here again. Accordingly, the related-art details mentioned in the present embodiment can also be applied to the second embodiment.
A seventh embodiment of the present application relates to an apparatus for determining content diversity based on sliding spectrum decomposition, comprising:
a memory for storing instructions for execution by one or more processors of the system, an
A processor, one of the processors of the system, is adapted to execute the instructions to implement any one of the possible methods of the second embodiment described above.
The second embodiment is a method embodiment corresponding to the present embodiment, and the present embodiment may be implemented in cooperation with the second embodiment. The related technical details mentioned in the second embodiment are still valid in this embodiment, and in order to reduce repetition, they are not described here again. Accordingly, the related-art details mentioned in the present embodiment can also be applied to the second embodiment.
An eighth embodiment of the present application relates to a computer storage medium encoded with a computer program, the computer readable medium having stored thereon instructions that, when executed on a computer, cause the computer to perform any one of the possible methods of the second embodiment described above.
The second embodiment is a method embodiment corresponding to the present embodiment, and the present embodiment may be implemented in cooperation with the second embodiment. The related technical details mentioned in the second embodiment are still valid in this embodiment, and in order to reduce repetition, they are not described here again. Accordingly, the related-art details mentioned in the present embodiment can also be applied to the second embodiment.
It should be noted that, each method embodiment of the present application may be implemented in software, hardware, firmware, or the like. Regardless of whether the application is implemented in software, hardware, or firmware, the instruction code may be stored in any type of computer-accessible memory (e.g., permanent or modifiable, volatile or non-volatile, solid or non-solid, fixed or removable media, etc.). Also, the Memory may be, for example, programmable array logic (Programmable Array Logic, abbreviated as "PAL"), random access Memory (RandomAccess Memory, abbreviated as "RAM"), programmable Read-Only Memory (Programmable Read Only Memory, abbreviated as "PROM"), read-Only Memory (ROM), electrically erasable programmable Read-Only Memory (Electrically Erasable Programmable ROM, abbreviated as "EEPROM"), magnetic disk, optical disk, digital versatile disk (Digital Versatile Disc, abbreviated as "DVD"), and the like.
It should be noted that, in the embodiments of the present application, each unit/module mentioned in each device is a logic unit/module, and in physical terms, one logic unit may be a physical unit, or may be a part of one physical unit, or may be implemented by a combination of multiple physical units, where the physical implementation manner of the logic unit itself is not the most important, and the combination of functions implemented by the logic units is the key to solve the technical problem set forth in the present application. Furthermore, in order to highlight the innovative part of the present application, the above-described device embodiments of the present application do not introduce elements that are less closely related to solving the technical problem presented by the present application, which does not indicate that other elements are not present in the above-described device embodiments.
It should be noted that in the claims and the description of this patent, relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
While the present application has been shown and described with reference to certain preferred embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present application.
Claims (8)
1. A recommendation method based on content diversity for an electronic device, the method comprising:
screening Z contents according to the user portrait;
and scoring each content of the Z contents according to a preset rule to obtain the correlation between the content and the user, wherein the correlation is shown in the following formula:
wherein T represents the T-th of the T contents;a relevance score representing a T-th of the T content;
n candidate contents are obtained from the Z contents according to the relevance ranking;
determining a diversity of any T pieces of content in the N pieces of candidate content, wherein the T pieces of content comprise notes from different categories;
determining a sliding window with the size w according to the display size of the electronic equipment or the number of notes which can be browsed or perceived by a user;
sliding the T contents by using the sliding window based on a time sequence to obtain a content matrix X, wherein the content matrix X comprises T contents expressed by a space vector vi of the contents, i represents the identification of the contents, and the value of i is a positive integer less than or equal to T;
singular value decomposition is carried out on the content matrix X to obtain a plurality of singular values, wherein the singular values are shown in the following formula:
And taking the product of the singular values as the diversity value of the T contents, as shown in the following formula:
wherein sigma ijk Representing the singular values; u (u) i (1) ∈R L ，u j (2) ∈R w ，u k (3) ∈R d Respectively representing columns of an orthogonal decomposition matrix of the content matrix X;representing an outer product operation;
selecting T contents with the largest sum of the relativity and the diversity from the N candidate contents, sequencing and recommending the T contents to the user, wherein the T contents are shown in the following formula:
wherein γ is a hyper-parameter for adjusting the balance of the correlation and the diversity.
2. The method of claim 1, wherein sliding the T content with the sliding window based on a time sequence results in a content matrix, comprising:
aggregating w contents in each sliding window into one row to obtain an intermediate matrix A L×w ；
Replacing the elements in the intermediate matrix with the space vectors vi to obtain a content matrix X epsilon R L×w×d ；
Wherein l=t-w+m; m represents the number of the contents sliding each time on the T contents of the sliding window; d represents the dimension of the spatial vector of the content.
3. The method of claim 1, wherein w is less than or equal to T.
4. The method of claim 1, wherein the sliding window slides m of the contents at a time over the T contents, wherein 1.ltoreq.m.ltoreq.w.
5. The method of claim 1, wherein determining a sliding window of size w based on a display size of the electronic device comprises:
taking the quantity of the content which can be displayed by the display screen of the electronic equipment as the size w of the sliding window; or alternatively
And expanding the number after a preset amplitude based on the number of the content which can be displayed by the display screen of the electronic equipment, and taking the number as the size w of the sliding window.
6. A content diversity-based recommendation apparatus, the apparatus comprising:
and a determination module:
screening Z contents according to the user portrait;
and scoring each content of the Z contents according to a preset rule to obtain the correlation between the content and the user, wherein the correlation is shown in the following formula:
wherein T represents the T-th of the T contents;a relevance score representing a T-th of the T content;
n candidate contents are obtained from the Z contents according to the relevance ranking;
determining a diversity of any T pieces of content in the N pieces of candidate content, wherein the T pieces of content comprise notes from different categories;
determining a sliding window with the size w according to the display size of the electronic equipment or the number of notes which can be browsed by a user or the number of notes which can be perceived by the user;
The processing module is used for:
sliding the T contents by using the sliding window based on a time sequence to obtain a content matrix X, wherein the content matrix X comprises the T contents represented by space vectors vi of the contents, i represents the identification of the contents, and the value of i is a positive integer less than or equal to T;
the calculation module:
singular value decomposition is carried out on the content matrix X to obtain a plurality of singular values, wherein the singular values are shown in the following formula:
and taking the product of the singular values as the diversity value of the T contents, as shown in the following formula:
wherein sigma ijk Representing the singular values; u (u) i (1) ∈R L ，u j (2) ∈R w ，u k (3) ∈R d Respectively representing columns of an orthogonal decomposition matrix of the content matrix X;representing an outer product operation;
selecting T contents with the largest sum of the relativity and the diversity from the N candidate contents, sequencing and recommending the T contents to the user, wherein the T contents are shown in the following formula:
wherein γ is a hyper-parameter for adjusting the balance of the correlation and the diversity.
7. A recommendation device based on content diversity, comprising:
a memory for storing instructions for execution by one or more processors of the system, and the processor, being one of the processors of the system, for executing the instructions to implement the content diversity-based recommendation method of any one of claims 1-5.
8. A computer readable storage medium encoded with a computer program, characterized in that the computer readable medium has stored thereon instructions, which when executed on a computer, cause the computer to perform the content diversity based recommendation method according to any of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110779586.6A CN113378063B (en) | 2021-07-09 | 2021-07-09 | Method for determining content diversity based on sliding spectrum decomposition and content sorting method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110779586.6A CN113378063B (en) | 2021-07-09 | 2021-07-09 | Method for determining content diversity based on sliding spectrum decomposition and content sorting method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113378063A CN113378063A (en) | 2021-09-10 |
| CN113378063B true CN113378063B (en) | 2023-07-28 |
Family
ID=77581609
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110779586.6A Active CN113378063B (en) | 2021-07-09 | 2021-07-09 | Method for determining content diversity based on sliding spectrum decomposition and content sorting method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113378063B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111724235A (en) * | 2020-06-09 | 2020-09-29 | 清华大学深圳国际研究生院 | Online commodity recommendation method based on user novelty |
| CN112270229A (en) * | 2020-10-16 | 2021-01-26 | 西安工程大学 | Landslide mass displacement prediction method based on singular spectrum analysis |
| CN112381351A (en) * | 2020-10-15 | 2021-02-19 | 广西电网有限责任公司 | Power utilization behavior change detection method and system based on singular spectrum analysis |
| CN112733006A (en) * | 2019-10-14 | 2021-04-30 | 中国移动通信集团上海有限公司 | User portrait generation method, device, equipment and storage medium |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10499176B2 (en) * | 2013-05-29 | 2019-12-03 | Qualcomm Incorporated | Identifying codebooks to use when coding spatial components of a sound field |
| KR101896273B1 (en) * | 2016-11-15 | 2018-09-10 | 금오공과대학교 산학협력단 | Noncontact bio-signal measurement method using SVD(Singular Value Decomposition) |
| CN109657646B (en) * | 2019-01-07 | 2023-04-07 | 哈尔滨工业大学（深圳） | Method and device for representing and extracting features of physiological time series and storage medium |
| CN111089726B (en) * | 2020-01-16 | 2021-12-03 | 东南大学 | Rolling bearing fault diagnosis method based on optimal dimension singular spectrum decomposition |
| CN111310034B (en) * | 2020-01-23 | 2023-04-07 | 深圳市雅阅科技有限公司 | Resource recommendation method and related equipment |
| CN111625710B (en) * | 2020-04-09 | 2021-12-24 | 北京百度网讯科技有限公司 | Processing method and device of recommended content, electronic equipment and readable storage medium |
| CN112697484A (en) * | 2020-11-09 | 2021-04-23 | 山东柯瑞申智能科技有限公司 | SSD multi-scale fluctuation analysis state monitoring method and device |
-
2021
- 2021-07-09 CN CN202110779586.6A patent/CN113378063B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112733006A (en) * | 2019-10-14 | 2021-04-30 | 中国移动通信集团上海有限公司 | User portrait generation method, device, equipment and storage medium |
| CN111724235A (en) * | 2020-06-09 | 2020-09-29 | 清华大学深圳国际研究生院 | Online commodity recommendation method based on user novelty |
| CN112381351A (en) * | 2020-10-15 | 2021-02-19 | 广西电网有限责任公司 | Power utilization behavior change detection method and system based on singular spectrum analysis |
| CN112270229A (en) * | 2020-10-16 | 2021-01-26 | 西安工程大学 | Landslide mass displacement prediction method based on singular spectrum analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113378063A (en) | 2021-09-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| RU2729956C2 (en) | Detecting objects from visual search requests | |
| CN110941740B (en) | Video recommendation method and computer-readable storage medium | |
| Kristan et al. | The eighth visual object tracking VOT2020 challenge results | |
| KR102122373B1 (en) | Method and apparatus for obtaining user portrait | |
| US11354720B2 (en) | Item recommendation techniques | |
| CN109697629B (en) | Product data pushing method and device, storage medium and computer equipment | |
| US20220122099A1 (en) | Analytical precursor mining for personalized recommendation | |
| US10049139B2 (en) | Diversity within search results | |
| US20190156395A1 (en) | System and Method for Analyzing and Searching for Features Associated with Objects | |
| US20170039198A1 (en) | Visual interactive search, scalable bandit-based visual interactive search and ranking for visual interactive search | |
| US20180181569A1 (en) | Visual category representation with diverse ranking | |
| US9177341B2 (en) | Determining search relevance from user feedback | |
| US20140067535A1 (en) | Concept-level User Intent Profile Extraction and Applications | |
| CN110008397B (en) | Recommendation model training method and device | |
| CN107590505B (en) | Learning method combining low-rank representation and sparse regression | |
| CN111400615B (en) | Resource recommendation method, device, equipment and storage medium | |
| CN110866805A (en) | Method and system for recommending object | |
| Dhillon et al. | Modeling dynamic user interests: A neural matrix factorization approach | |
| Hazrati et al. | Addressing the New Item problem in video recommender systems by incorporation of visual features with restricted Boltzmann machines | |
| CN101957825A (en) | Method for searching image based on image and video content in webpage | |
| US8903817B1 (en) | Determining search relevance from user feedback | |
| CN113378063B (en) | Method for determining content diversity based on sliding spectrum decomposition and content sorting method | |
| Nguyen et al. | Cold-start problems in recommendation systems via contextual-bandit algorithms | |
| Zahrawi et al. | Implementing recommender systems using machine learning and knowledge discovery tools | |
| CN113378065B (en) | Method for determining content diversity based on sliding spectrum decomposition and method for selecting content |
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 |