AU2020335019B2 - Evaluation method based on mobile news client and system thereof - Google Patents

Evaluation method based on mobile news client and system thereof Download PDF

Info

Publication number
AU2020335019B2
AU2020335019B2 AU2020335019A AU2020335019A AU2020335019B2 AU 2020335019 B2 AU2020335019 B2 AU 2020335019B2 AU 2020335019 A AU2020335019 A AU 2020335019A AU 2020335019 A AU2020335019 A AU 2020335019A AU 2020335019 B2 AU2020335019 B2 AU 2020335019B2
Authority
AU
Australia
Prior art keywords
content
recommended
evaluation
bad
title
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
Application number
AU2020335019A
Other versions
AU2020335019A1 (en
Inventor
Feng Cao
Xiaofei Dong
Lin Shi
Mingjun SUN
Dan Zhang
Tao Zhang
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing New Generation Artificial Intelligence Research Institute Co Ltd
Original Assignee
Nanjing New Generation Artificial Intelligence Research Institute Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nanjing New Generation Artificial Intelligence Research Institute Co Ltd filed Critical Nanjing New Generation Artificial Intelligence Research Institute Co Ltd
Publication of AU2020335019A1 publication Critical patent/AU2020335019A1/en
Application granted granted Critical
Publication of AU2020335019B2 publication Critical patent/AU2020335019B2/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis

Abstract

The disclosure discloses an evaluation method based on a mobile news client and system thereof, and relates to the field of Internet news information operation and service platforms, the method comprising: step 1: recommended content acquisition: simulating a user portrait to request data from a server, collecting pushed content returned by the server and storing the pushed content in a database; step 2: evaluation on the recommended content in two dimensions of quality and aging; step 3: feeding back a final result of the evaluation. Compared with the prior art, personalized recommended content evaluation is carried out from the APP side of the mobile news client, and it can be simulated to collect the content recommended by different users in real time according to different user portraits. The credibility of the content is evaluated in two dimensions of aging and quality, and the real-time performance, the information coverage and the quality of the content of the news information can be well reflected. Figure 2 is designated as the drawing for the abstract

Description

Evaluation Method Based on Mobile News Client and System Thereof
Technical Field The disclosure belongs to the field of Internet news information operation and service platforms, and particularly relates to an evaluation method based on a mobile news client and a system thereof
Background Art In recent years, with the development of artificial intelligence technology, personalized recommendation algorithms have been widely used in news push, live web, small video push, goods and other scenarios. However, at present, there are many problems in the process of recommending content only by means of recommendation algorithm, especially in the field of news information recommendation, such as lack of values, information cocoon houses, narrow information and so on. In order to prevent the Internet news information operation and service platforms from becoming the help of bad content dissemination, we shall be aware that the algorithm determines the content and to be aware of being trapped in the information cocoon house by the algorithm, and it is necessary to evaluate the content of the news recommendation system in order to ensure the credibility of the dissemination content of the news recommendation system. The prior art solution is as follows: a content collection method: creating the same number of threads according to a number of the obtained news clients, obtaining a number of cores of the central processing unit (CPU) of the system and binding each thread on the core of the corresponding CPU according to a preset rule; storing data in all preset analysis queues on the same core of CPU in corresponding preset output queues, and when an output instruction is received, transmitting the data in the preset output queues to a preset database, so that supervision on news client data is realized on the basis of the data in the preset database. content evaluation method: evaluation metrics for recommendation systems typically range a number of dimensions: user experience, algorithm accuracy, and business goals. There are many problems with this approach. For example: regarding the content obtaining method, through directly obtaining the column content of each news client, it is difficult to evaluate the recommended content of each user for the recommendation system; and regarding the content evaluation, the current evaluation index is based on the general recommendation algorithm, and there is no specific evaluation index for news information recommendation system.
Summary of the Disclosure Aiming at a plurality of problems existing in the prior art, the disclosure aims to propose an evaluation method based on a mobile news client and a system thereof, which can evaluate the credibility of the recommended content from a plurality of dimensions by acquiring personalized recommended content of the mobile news client and analyzing the recommended content in real time. In order to achieve the objects, the technical solution adopted by the disclosure is as follows: an evaluation method based on a mobile news client, comprising: step 1: recommended content acquisition: simulating a user portrait to request data from a server, collecting pushed content returned by the server and storing the pushed content in a database; step 2: evaluation on the recommended content in two dimensions of content quality and aging; step 3: feeding back a final result of the evaluation. Further, step 1 specifically comprises: firstly, presetting a user operation instruction, comprising: refreshing, reading and clicking a like, then simulating the user operation instruction, transmitting the user operation instruction to a virtualized news client in an adb mode, the news client requesting data from the server through simulated clicking according to the instruction, the server returning pushed personalized content to the news client, obtaining the recommended content from the news client using a hook server, and storing the recommended content in the database. Further, evaluation on the content quality in step 2 specifically comprises: firstly, identifying bad content: constructing a large amount of training data for marking words of the bad content adopting a sequence labeling method, and carrying out model training; predicting obtained recommended data using a trained model, judging whether the words of the bad content are contained, extracting and recording the words of the bad content if the words of the bad content are contained, and identifying the content as a Title Party content if the words of the bad content are not contained; identifying the content as the Title Party content: constructing a large amount of training data adopting a two-classification model, inputting a title and a text of an article together as a model, marking the title to be o consistent with an article as 1 and marking the title to be inconsistent with the article as 0, and performing classification model training; inputting the title and the text of the article of the obtained recommended data into the model, and judging whether the content is the Title Party content or not; evaluation on the aging in step 2 specifically comprises: evaluating real-time performance of the system through a content recommendation rate in last n days, taking n=3, that is, a ratio of the recommended quantity within a production time of the recommended content of the latest 72 hours to the total recommended quantity, obtaining the production time of the recommended content, and calculating data proportion within the production time of 72 hours. Further, an aging index and a quality index obtained in step 2 are calculated, and an average value is taken as the final result. The disclosure also provides a system for implementing the evaluation method, comprising a recommended content acquisition module, a content credibility evaluation module and an evaluation result feedback module; the recommended content acquisition module comprising: a user portrait simulation and data request submodule: simulating browsing, clicking and sharing behaviors of a user to form a preset user portrait, and requesting data from the news client to a server according to the preset user portrait; a server pushing submodule: the server pushing personalized recommended news information content to the news client; a content collection submodule: acquiring the recommended content from the news client using a hook server; a data storage submodule: storing a collected content in a database; the content credibility evaluation module comprising: an aging evaluation submodule: evaluating real-time performance of a recommendation system through a news recommendation rate in last n days, wherein N <= 3; a quality evaluation submodule: detecting whether the recommended content contains bad content or Title Party content. The disclosure also provides an evaluation method based on a mobile news client. The evaluation method comprises: step 1: recommended content acquisition: simulating a user portrait to request data from a server, collecting pushed content returned by the server and storing the pushed content in a database; step 2: evaluation on the recommended content in two dimensions of content quality and aging; step 3: feeding back a final result of the evaluation, wherein evaluation on the content quality in step 2 specifically comprises: firstly, identifying bad content: constructing a large amount of training data for marking words of the bad content adopting a sequence labeling method, and carrying out model training; predicting obtained recommended data using a trained model, judging whether the words of the bad content are contained, extracting and recording the words of the bad content if the words of the bad content are contained, and identifying the content as a Title Party content if the words of the bad content are not contained; identifying the content as the Title Party content: constructing a large amount of training data adopting a two-classification model, inputting a title and a text of an article together as a model, marking the title to be consistent with an article as 1 and marking the title to be inconsistent with the article as 0, and performing classification model training; inputting the title and the text of the article of the obtained recommended data into the model, and judging whether the content is the Title Party content or not; and finally calculating a quality index of the content, wherein the formula is as follows:
LiEu ALLi S= U
wherein: Q is a ratio of a quantity of the recommended content, detected at a moment, which does not meet a content quality requirement to a total recommended quantity, U is a total quantity of simulated different user portraits, qi is the quantity of the recommended content that user portrait i acquires which does not meet the content quality requirement, and ALLi is a total quantity of the recommended content that user portrait i acquires. The disclosure also provide an evaluation system of a news client. The evaluation system comprises a recommended content acquisition module, a content credibility evaluation module and an evaluation result feedback module; the recommended content acquisition module comprising: a user portrait simulation and data request submodule: simulating browsing, clicking and sharing behaviors of a user to form a preset user portrait, and requesting data from the news client to a server according to the preset user portrait; a server pushing submodule: the server pushing personalized recommended news information content to the news client; a content collection submodule: acquiring the recommended content from the news client using a hook server; a data storage submodule: storing a collected content in a database; the content credibility evaluation module comprising: an aging evaluation submodule: evaluating real-time performance of a recommendation system through a news recommendation rate in last n days, wherein N <= 3; a quality evaluation submodule: detecting whether the recommended content contains bad content or Title Party content: firstly, identifying bad content: constructing a large amount of training data for marking words of the bad content adopting a sequence labeling method, and carrying out model training; predicting obtained recommended data using a trained model, judging whether the words of the bad content are contained, extracting and recording the words of the bad content if the words of the bad content are contained, and identifying the content as a Title Party content if the words of the bad content are not contained; identifying the content as the Title Party content: constructing a large amount of training data adopting a two-classification model, inputting a title and a text of an article together as a model, marking the title to be consistent with an article as 1 and marking the title to be inconsistent with the article as 0, and performing classification model training; inputting the title and the text of the article of the obtained recommended data into the model, and judging whether the content is the Title Party content or not; and finally calculating a quality index of the content, wherein the formula is as follows:
Lieu ALLi S= U
wherein: Q is a ratio of a quantity of the recommended content, detected at a moment, which does not meet a content quality requirement to a total recommended quantity, U is a total quantity of simulated different user portraits, qi is the quantity of the recommended content that user portrait i acquires which does not meet the content quality requirement, and ALLi is a total quantity of the recommended content that user portrait i acquires; and the evaluation result feedback module evaluating data obtained by simulating different user portraits at each moment to obtain an aging index and a quality index, and taking an average value as a final result and feeding back the final result to a monitoring platform. Compared with the prior art, the present disclosure has the following beneficial effects: (1) in the aspect of content acquisition, personalized recommended content acquisition is carried out from the APP side of the mobile news client, and content recommended by different users according to different user portraits can be simulated to be acquired in real time; (2) in the aspect of content evaluation, the credibility of the content is evaluated in two dimensions of aging and quality, and the real-time performance, the information coverage and the quality of the content of news information can be well reflected; and (3) the evaluation result and the detected bad content are fed back in real time, so that it is convenient to manage the platform.
Brief Description of the Drawings Figure 1 is a schematic diagram of a collection method of news client data in prior art; Figure 2 is a schematic diagram of an evaluation system of a mobile news client of the present disclosure; Figure 3 is a schematic diagram of a recommended content acquisition module in the evaluation system of the present disclosure; Figure 4 is a flowchart of an collection method for the recommended content in the evaluation method of the present disclosure; and Figure 5 is a flowchart of a content evaluation method in the evaluation method of the present disclosure.
Detailed Description of the Disclosure In order that the present disclosure may be better understood by those skilled in the art, the present disclosure will now be described in detail with reference to specific embodiments. Figure 1 shows a schematic diagram of a collection method of news client data in prior art, and the collection method comprises creating the same number of threads according to a number of the obtained news clients, obtaining a number of cores of the central processing unit (CPU) of the system and binding each thread on the core of the corresponding CPU according to a preset rule; storing data in all preset analysis queues on the same core of CPU in corresponding preset output queues, and when an output instruction is received, transmitting the data in the preset output queues to a preset database, so that supervision on news client data is realized on the basis of the data in the preset database. In this collection method, through directly obtaining the column content of each news client, it is difficult to evaluate the recommended content of each user for the recommendation system. As shown in figure 2, the disclosure provides an evaluation method based on a mobile news client, comprising: step 1: recommended content acquisition: simulating a user portrait to request data from a server, collecting pushed content returned by the server and storing the pushed content in a database; step 2: evaluation on the recommended content in two dimensions of quality and aging; step 3: feeding back a final result of the evaluation. As shown in figures 3 and 4, the recommended content acquisition method in the evaluation method of the present disclosure is specifically as follows: firstly, presetting a user operation instruction, comprising: refreshing, reading and clicking a like, then simulating the user operation instruction, transmitting the user operation instruction to a virtualized news client in an adb mode, the news client requesting data from the server through simulated clicking according to the instruction, the server returning pushed personalized content to the news client, obtaining the recommended content from the news client using a hook server, and storing the recommended content in the database. As shown in figure 5, in the flow chart of the content evaluation method in the evaluation method provided by the present disclosure, the specific approach of evaluation on the content quantity is as follows: firstly, identifying bad content: constructing a large amount of training data for marking words of the bad content adopting a sequence labeling method, and carrying out model training; predicting obtained recommended data using a trained model, judging whether the words of the bad content are contained, extracting and recording the words of the bad content if the words of the bad content are contained, and identifying the content as a Title Party content if the words of the bad content are not contained; identifying the content as the Title Party content: constructing a large amount of training data adopting a two-classification model, inputting a title and a text of an article together as a model, marking the title to be consistent with an article as 1 and marking the title to be inconsistent with the article as 0, and performing classification model training; inputting the title and the text of the article of the obtained recommended data into the model, and judging whether the content is the Title Party content or not; and finally calculating a quality index of the content, wherein the formula is as follows:
Q = ALL U Formula (1)
wherein: Q is a ratio of a quantity of the recommended content, detected at a moment, which does not meet a content quality requirement to a total recommended quantity, U is a total quantity of simulated different user portraits, qi is the quantity of the recommended content that user portrait i acquires which does not meet the content quality requirement, and ALLi is a total quantity of the recommended content that user portrait i acquires. The specific approach of the evaluation on the aging is as follows: evaluating real-time performance of the system through a content recommendation rate in last n days, taking n=3, that is, a ratio of the recommended quantity within a production time of the recommended content of the latest 72 hours to the total recommended quantity, obtaining the production time of the recommended content, and calculating data proportion within the production time of 72 hours, wherein the formula is as follows:
T = LL Formula (2) U wherein T is a ratio of a quantity of the recommended content with the production time of 72 hours to the total recommended content, U is a total quantity of simulated different user portraits, ti is a quantity of the recommended content that user portrait i acquires which does not meet an aging requirement, and ALLi is a total quantity of the recommended content that user portrait i acquires. In step 3, data obtained by simulating different user portraits at each moment are subjected to the evaluation method to obtain an aging index and a quality index, and an average value is taken as the final result and fed back to a monitoring platform. The disclosure also provides a system for implementing the evaluation and monitoring method based on a news client, comprising a recommended content acquisition module, a content credibility evaluation module and an evaluation result feedback module; the recommended content acquisition module comprising: a user portrait simulation and data request submodule: simulating browsing, clicking and sharing behaviors of a user to form a preset user portrait, and requesting data from the news client to a server according to the preset user portrait; a server pushing submodule: the server pushing personalized recommended news information content to the news client; a content collection submodule: acquiring the recommended content from the news client using a hook server; a data storage submodule: storing a collected content in a database; the content credibility evaluation module comprising: an aging evaluation submodule: evaluating real-time performance of a recommendation system through a news recommendation rate in last n days, wherein N < 3; a quality evaluation submodule: detecting whether the recommended content contains bad content or Title Party content. According to the evaluation method based on a news client, in the aspect of content collection, personalized recommended content collection is carried out from the APP side of the mobile news client, content recommended by different users according to different user portraits can be simulated to be collected in real time, and recommended content of users of the same type (users with similar user portraits) does not need to be collected completely. The evaluation method is more flexible, and the scale of the collected content can be controlled freely. The scale of collected personalized content can be adjusted freely according to relevant rules. o In the aspect of content evaluation, the credibility of recommended content on news information is evaluated in two dimensions of aging and quality, and the real-time performance of the news information and the quality of article content can be well reflected. The embodiments described above are merely illustrative of the technical idea of the present disclosure and should not be taken as limiting the scope of the present disclosure, but any changes made on the basis of the technical solution according to the technical idea proposed by the present disclosure fall within the scope of the claims of the present disclosure. The techniques not involved in the present disclosure can be implemented with the prior art. Throughout this specification and the claims that follow unless the context requires otherwise, the words 'comprise' and 'include' and variations such as 'comprising' and 'including' will be understood to imply the inclusion of a stated integer or group of integers but not the exclusion of any other integer or group of integers. The reference to any prior art in this specification is not, and should not be taken as, an acknowledgment or any form of suggestion that such prior art forms part of the common general knowledge of the technical field.

Claims (5)

  1. Claims 1. An evaluation method based on a mobile news client, comprising: step 1: recommended content acquisition: simulating a user portrait to request data from a server, collecting pushed content returned by the server and storing the pushed content in a database; step 2: evaluation on the recommended content in two dimensions of content quality and aging; step 3: feeding back a final result of the evaluation, wherein evaluation on the content quality in step 2 specifically comprises: firstly, identifying bad content: constructing a large amount of training data for marking words of the bad content adopting a sequence labeling method, and carrying out model training; predicting obtained recommended data using a trained model, judging whether the words of the bad content are contained, extracting and recording the words of the bad content if the words of the bad content are contained, and identifying the content as a Title Party content if the words of the bad content are not contained; identifying the content as the Title Party content: constructing a large amount of training data adopting a two-classification model, inputting a title and a text of an article together as a model, marking the title to be consistent with an article as 1 and marking the title to be inconsistent with the article as 0, and performing classification model training; inputting the title and the text of the article of the obtained recommended data into the model, and judging whether the content is the Title Party content or not; and finally calculating a quality index of the content, wherein the formula is as follows:
  2. LiEu ALLi S= U wherein: Q is a ratio of a quantity of the recommended content, detected at a moment, which does not meet a content quality requirement to a total recommended quantity, U is a total quantity of simulated different user portraits, qi is the quantity of the recommended content that user portrait i acquires which does not meet the content quality requirement, and ALLi is a total quantity of the recommended content that user portrait i acquires. 2. The evaluation method based on a mobile news client according to claim 1, wherein step 1 specifically comprises: firstly, presetting a user operation instruction, comprising: refreshing, reading and clicking a like, then simulating the user operation instruction, transmitting the user operation instruction to a virtualized news client in an adb mode, the news client requesting data from the server through simulated clicking according to the instruction, the server returning pushed personalized content to the news client, obtaining the recommended content from the news client using a hook server, and storing the recommended content in the database.
  3. 3. The evaluation method based on a mobile news client according to claim 1, wherein evaluation on the aging in step 2 specifically comprises: evaluating real-time performance of the system through a content recommendation rate in last n days, taking n=3, that is, a ratio of the recommended quantity within a production time of the recommended content of the latest 72 hours to the total recommended quantity, obtaining the production time of the recommended content, and calculating data proportion within the production time of 72 hours, wherein the formula is as follows: t EiuALL T = U wherein T is a ratio of a quantity of the recommended content with the production time of 72 hours to the total recommended content, U is a total quantity of simulated different user portraits, ti is a quantity of the recommended content that user portrait i acquires which does not meet an aging requirement, and ALLi is a total quantity of the recommended content that user portrait i acquires.
  4. 4. The evaluation method based on a mobile news client according to claim 2, wherein step 3 specifically comprises: calculating an aging index and a quality index obtained in step 2, and taking an average value as the final result.
  5. 5. An evaluation system of a news client, comprising a recommended content acquisition module, a content credibility evaluation module and an evaluation result feedback module; the recommended content acquisition module comprising: a user portrait simulation and data request submodule: simulating browsing, clicking and sharing behaviors of a user to form a preset user portrait, and requesting data from the news client to a server according to the preset user portrait; a server pushing submodule: the server pushing personalized recommended news information content to the news client; a content collection submodule: acquiring the recommended content from the news client using a hook server; a data storage submodule: storing a collected content in a database; the content credibility evaluation module comprising: an aging evaluation submodule: evaluating real-time performance of a recommendation system through a news recommendation rate in last n days, wherein N <= 3; a quality evaluation submodule: detecting whether the recommended content contains bad content or Title Party content: firstly, identifying bad content: constructing a large amount of training data for marking words of the bad content adopting a sequence labeling method, and carrying out model training; predicting obtained recommended data using a trained model, judging whether the words of the bad content are contained, extracting and recording the words of the bad content if the words of the bad content are contained, and identifying the content as a Title Party content if the words of the bad content are not contained; identifying the content as the Title Party content: constructing a large amount of training data adopting a two-classification model, inputting a title and a text of an article together as a model, marking the title to be consistent with an article as 1 and marking the title to be inconsistent with the article as 0, and performing classification model training; inputting the title and the text of the article of the obtained recommended data into the model, and judging whether the content is the Title Party content or not; and finally calculating a quality index of the content, wherein the formula is as follows:
    Q LieuALLi S= U wherein: Q is a ratio of a quantity of the recommended content, detected at a moment, which does not meet a content quality requirement to a total recommended quantity, U is a total quantity of simulated different user portraits, qi is the quantity of the recommended content that user portrait i acquires which does not meet the content quality requirement, and ALLi is a total quantity of the recommended content that user portrait i acquires; and the evaluation result feedback module evaluating data obtained by simulating different user portraits at each moment to obtain an aging index and a quality index, and taking an average value as a final result and feeding back the final result to a monitoring platform.
    Creating Generating Analyzing thread acquisition text Database to bind the and information storage thread with collection of the news client CPU node
    Fig. 1
    User portrait Recommended content acquisition
    Content credibility evaluation simulation and data request
    Quality Server evaluation pushing Evaluation result feedback Content Aging collection evaluation
    Data storage
    Fig. 2
    Control Virtualized section section
    User portrait simulation and request News client
    Data storage Content collection
    Fig. 3
    Start
    Inputting user operation
    Controlling the client in an adb mode
    Simulating clicking on the client
    Content collection
    Storing to database
    End
    Fig. 4
    Start
    Inputting the recommended content Containing the words Acquiring the of the bad content production time of the Identifying the words recommended content of the bad content
    Containing no words of the bad content Yes Judging whether the Judging whether the production time is content is the Title Yes less than 72 hours Party content or not
    No No Evaluated content Evaluated content meeting the aging Non real-time meeting the quality Title Party Bad content requirement content requirement content
    Feeding back the Calculating the Calculating the evaluation result aging index quality index
    Fig. 5
AU2020335019A 2019-12-31 2020-03-31 Evaluation method based on mobile news client and system thereof Active AU2020335019B2 (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
CN201911409113.6A CN111143688B (en) 2019-12-31 2019-12-31 Evaluation method and system based on mobile news client
CN201911409113.6 2019-12-31
PCT/CN2020/082285 WO2021134944A1 (en) 2019-12-31 2020-03-31 Mobile news client-based evaluation method and system therefor

Publications (2)

Publication Number Publication Date
AU2020335019A1 AU2020335019A1 (en) 2021-07-15
AU2020335019B2 true AU2020335019B2 (en) 2022-06-16

Family

ID=70522566

Family Applications (1)

Application Number Title Priority Date Filing Date
AU2020335019A Active AU2020335019B2 (en) 2019-12-31 2020-03-31 Evaluation method based on mobile news client and system thereof

Country Status (3)

Country Link
CN (1) CN111143688B (en)
AU (1) AU2020335019B2 (en)
WO (1) WO2021134944A1 (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117291690A (en) * 2023-11-23 2023-12-26 深圳市万特网络科技有限公司 Intelligent product sales recommendation method based on data analysis

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150046425A1 (en) * 2013-08-06 2015-02-12 Hsiu-Ping Lin Methods and systems for searching software applications

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8286176B1 (en) * 2008-09-29 2012-10-09 Amazon Technologies, Inc. Optimizing resource configurations
CN103365902B (en) * 2012-03-31 2017-06-20 北大方正集团有限公司 The appraisal procedure and device of internet news
US9992690B2 (en) * 2013-10-11 2018-06-05 Textron Innovations, Inc. Placed wireless instruments for predicting quality of service
CN107818156A (en) * 2017-10-31 2018-03-20 广东思域信息科技有限公司 A kind of real time individual news recommends method and system
CN107968842B (en) * 2017-12-26 2020-12-08 百度在线网络技术（北京）有限公司 News pushing method, device and equipment based on distributed system
CN109684582A (en) * 2018-11-08 2019-04-26 张耀伦 A kind of evaluating system and method for information resources
CN110413890A (en) * 2019-07-29 2019-11-05 武汉匠楚科技有限公司 A kind of method that news recommender system polymerization news is presented

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150046425A1 (en) * 2013-08-06 2015-02-12 Hsiu-Ping Lin Methods and systems for searching software applications

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
BASHIR, M. A. et al., ""Recommended For You" A First Look at Content Recommendation Networks", Proceedings of the Internet Measurement Conference (2016), pages 17-24, published 14 November 2016 *

Also Published As

Publication number Publication date
CN111143688A (en) 2020-05-12
WO2021134944A1 (en) 2021-07-08
CN111143688B (en) 2021-03-02
AU2020335019A1 (en) 2021-07-15

Similar Documents

Publication Publication Date Title
CN109902708B (en) Recommendation model training method and related device
CN108921221B (en) User feature generation method, device, equipment and storage medium
US11809450B2 (en) Selectively identifying and recommending digital content items for synchronization
US8682830B2 (en) Information processing apparatus, information processing method, and program
CN112153426B (en) Content account management method and device, computer equipment and storage medium
WO2021129055A1 (en) Information prediction model training method and apparatus, information prediction method and apparatus, storage medium, and device
CN110543598A (en) information recommendation method and device and terminal
CN115238173B (en) Behavior analysis and medical service pushing method, equipment and medium based on big data
CN110019616A (en) A kind of POI trend of the times state acquiring method and its equipment, storage medium, server
CN105573995A (en) Interest identification method, interest identification equipment and data analysis method
KR20210066754A (en) Method of generating deep learning model for user characteristic analysis using federated learning
CN111858969B (en) Multimedia data recommendation method, device, computer equipment and storage medium
CN112100221B (en) Information recommendation method and device, recommendation server and storage medium
CN113312531A (en) User portrait identification method based on DPI analysis and decision tree model
CN115718831A (en) Social media robot detection method based on dynamic heterogeneous graph and related device
AU2020335019B2 (en) Evaluation method based on mobile news client and system thereof
CN113656699B (en) User feature vector determining method, related equipment and medium
CN107547626B (en) User portrait sharing method and device
CN114443671A (en) Recommendation model updating method and device, computer equipment and storage medium
CN117112902A (en) Off-line reinforcement learning commodity recommendation system based on multi-mode contrast learning
CN112565902A (en) Video recommendation method and device and electronic equipment
CN113962417A (en) Video processing method and device, electronic equipment and storage medium
CN116303663A (en) User affinity calculation method and system based on content social platform
WO2023087933A1 (en) Content recommendation method and apparatus, device, storage medium, and program product
CN113297417B (en) Video pushing method, device, electronic equipment and storage medium

Legal Events

Date Code Title Description
FGA Letters patent sealed or granted (standard patent)