CN110175264A - Construction method, server and the computer readable storage medium of video user portrait - Google Patents
Construction method, server and the computer readable storage medium of video user portrait Download PDFInfo
- Publication number
- CN110175264A CN110175264A CN201910332502.7A CN201910332502A CN110175264A CN 110175264 A CN110175264 A CN 110175264A CN 201910332502 A CN201910332502 A CN 201910332502A CN 110175264 A CN110175264 A CN 110175264A
- Authority
- CN
- China
- Prior art keywords
- video
- keyword
- user
- portrait
- target user
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/73—Querying
- G06F16/735—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/78—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/7867—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using information manually generated, e.g. tags, keywords, comments, title and artist information, manually generated time, location and usage information, user ratings
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Library & Information Science (AREA)
- Computational Linguistics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of construction methods of video user portrait, it include: the historical operation video information for obtaining target user, and the corresponding keyword of the historical operation video information, it is that the target user constructs portrait according to the historical operation video information and the keyword.The invention also discloses a kind of server and computer readable storage mediums.In the present invention, the feature extracted from historical operation video information careful can portray target user's behavioral characteristic, the corresponding keyword of historical operation video information reflects target user's behavioral characteristic on the whole simultaneously, so that user's portrait of building both has the characteristics of general outline, again can careful embodiment user the characteristics of, thus provide a kind of construction method of more accurate user's portrait.
Description
Technical field
The present invention relates to the construction methods of field of computer technology more particularly to a kind of video user portrait, video user
The construction device and computer readable storage medium of portrait.
Background technique
The essence of building user's portrait is to extract user characteristics, is mainly visited using the user's history of storage on the server
It asks data, i.e., is analyzed and excavated using the mass data of massive logs and lane database, labeled " " to user, and " mark
Label " are the marks that can indicate a certain dimensional characteristics of user.
Currently when constructing user's portrait, it is normally based in the history access data analysis regular period of user and goes out user
Action trail feature, such feature lack to user behavior deeply and it is careful portrays, cause the later period to be based on user's portrait
Recommendation and data analysis result inaccuracy, or even influence user experience.
Above content is only used to facilitate the understanding of the technical scheme, and is not represented and is recognized that above content is existing skill
Art.
Summary of the invention
The main purpose of the present invention is to provide a kind of construction methods of video user portrait, server and computer-readable
Storage medium, it is intended to which solving the historical behavior track characteristic based on user in the prior art is that user constructs portrait, and portrait lacks
User behavior is deeply portrayed with careful, so as to cause later period recommendation and the technical problem of data analysis result inaccuracy.
To achieve the above object, the present invention provides a kind of construction method of video user portrait, the video user portrait
Construction method include the following steps:
Obtain the historical operation video information and the corresponding keyword of the historical operation video information of target user;
It is that the target user constructs portrait according to the historical operation video information and the keyword.
Preferably, described the step of obtaining the historical operation video information corresponding keyword, includes:
The associated video of the target user is determined according to the historical operation video information;
The corresponding keyword of the historical operation information is obtained according to the keyword of pre- setting video and the associated video.
Preferably, the construction method of the video user portrait is further comprising the steps of:
Obtain the description text of each pre- setting video;
Obtain the corresponding multiple crucial term vectors of description text of the pre- setting video;
Obtain the weight of each crucial term vector;
The keyword of the pre- setting video is determined from each keyword according to the weight.
Preferably, the step of weight for obtaining each crucial term vector includes:
Obtain first hit-count of the crucial term vector in the description text of corresponding pre- setting video;
Obtain the second hit-count in the description text of pre- setting video of the crucial term vector in preset quantity;
The weight of the crucial term vector is determined according to first hit-count and second hit-count.
Preferably, described that the crucial term vector is determined according to first hit-count and second hit-count
After the step of weight further include:
Judge whether the crucial term vector is located at the title or name entity of the description text of corresponding pre- setting video;
When the keyword vector is located at the title or name entity of the description text of corresponding pre- setting video, according to institute
The first modifying factor for stating weight generates revised weight;
When the crucial term vector is not located at the title or name entity of the description text of the pre- setting video, according to institute
The second modifying factor for stating weight generates revised weight, wherein first modifying factor is greater than second modifying factor
Son.
Preferably, the associated video includes that the video that browsed of the target user, the target user watched
The predetermined registration operation video of video and the target user, it is described that institute is obtained according to the keyword and the associated video of pre- setting video
The step of stating historical operation information corresponding keyword include:
The keyword of the associated video is obtained according to the keyword of the pre- setting video and the associated video;
The keyword of each associated video is obtained in the corresponding keyword of video that the target user browsed
The first hit-count;
The keyword of each associated video is obtained in the corresponding keyword of video that the target user watched
The second hit-count;
Obtain predetermined registration operation video corresponding keyword of the keyword in the target user of each associated video
In third hit-count;
According to first hit-count, second hit-count and the third hit-count from the associated video
Keyword in determine the corresponding keyword of the historical operation information.
Preferably, described that picture is constructed for the target user according to the historical operation video information and the keyword
As the step of include:
Obtain the historical operation video information of the user except the target user;
It is grasped according to the history of the user except the historical operation video information of the target user and the target user
The scoring of each associated video is obtained as video information;
The interest video of the target user is determined according to the scoring of each associated video;
It is that the target user constructs portrait according to the interest video and the keyword.
Preferably, the construction method of the video user portrait is further comprising the steps of:
The video historical operation information of user is updated in real time or periodically.
In addition, to achieve the above object, the present invention also provides a kind of server, which includes: memory, processor
And it is stored in the building processing routine for the video user portrait that can be run on the memory and on the processor, the view
The building processing routine of frequency user portrait realizes the building side of video user portrait as described above when being executed by the processor
The step of method.
In addition, to achieve the above object, the present invention also proposes a kind of computer readable storage medium, which is characterized in that institute
State the building processing routine that video user portrait is stored on computer readable storage medium, the building of the video user portrait
The step of construction method of video user portrait as described above is realized when processing routine is executed by processor.
Construction method, server and the readable computer storage for a kind of video user portrait that the embodiment of the present invention proposes are situated between
Matter obtains the historical operation video information and the corresponding keyword of the historical operation video information of target user, according to institute
It states historical operation video information and the keyword is that the target user constructs portrait.The invention also discloses a kind of services
Device and computer readable storage medium.In the present invention, the feature extracted from historical operation video information careful can portray
Target user's behavioral characteristic, while the corresponding keyword of historical operation video information reflects target user behavior spy on the whole
Point so that user's portrait of building not only has the characteristics of general outline, but also can be careful embodiment user the characteristics of, thus provide
A kind of construction method of more accurate user's portrait.
Detailed description of the invention
Fig. 1 is the terminal structure schematic diagram for the hardware running environment that the embodiment of the present invention is related to;
Fig. 2 is the flow diagram of the construction method first embodiment of video user of the present invention portrait;
Fig. 3 is the flow diagram of the construction method second embodiment of video user of the present invention portrait;
Fig. 4 is the flow diagram of the construction method 3rd embodiment of video user of the present invention portrait;
Fig. 5 is the flow diagram of the construction method fourth embodiment of video user of the present invention portrait.
The embodiments will be further described with reference to the accompanying drawings for the realization, the function and the advantages of the object of the present invention.
Specific embodiment
It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not intended to limit the present invention.
The primary solutions of the embodiment of the present invention are: obtaining the historical operation video information of target user and described
The corresponding keyword of historical operation video information is the target according to the historical operation video information and the keyword
User constructs portrait.
In the present invention, it is special that the feature extracted from historical operation video information careful can portray target user's behavior
Point, while the corresponding keyword of historical operation video information reflects target user's behavioral characteristic on the whole, so that building
User's portrait not only has the characteristics of general outline, but can be careful embodiment user the characteristics of, thus provide a kind of more accurate
User portrait construction method.
As shown in Figure 1, Fig. 1 is the terminal structure schematic diagram for the hardware running environment that the embodiment of the present invention is related to.
The terminal of that embodiment of the invention is server.
As shown in Figure 1, the server may include: processor 1001, such as CPU, communication bus 1002, memory
1003.Wherein, communication bus 1002 is for realizing the connection communication between these components.Memory 1005 can be high-speed RAM
Memory is also possible to stable memory (non-volatile memory), such as magnetic disk storage.Memory 1003 can
The storage device that can also be independently of aforementioned processor 1001 of choosing.
It will be understood by those skilled in the art that the restriction of the not structure paired terminal of terminal structure shown in Fig. 1, can wrap
It includes than illustrating more or fewer components, perhaps combines certain components or different component layouts.
As shown in Figure 1, as may include operating system and video in a kind of memory 1003 of computer storage medium
The building processing routine of user's portrait.
In device shown in Fig. 1, processor 1001 can be used for that the video user stored in memory 1003 is called to draw
The building processing routine of picture, and execute following operation:
Further, the building processing journey that processor 1001 can call the video user stored in memory 1003 to draw a portrait
Sequence also executes following operation:
Obtain the historical operation video information and the corresponding keyword of the historical operation video information of target user;
It is that the target user constructs portrait according to the historical operation video information and the keyword.
Further, the building processing journey that processor 1001 can call the video user stored in memory 1003 to draw a portrait
Sequence also executes following operation:
The associated video of the target user is determined according to the historical operation video information;
The corresponding keyword of the historical operation information is obtained according to the keyword of pre- setting video and the associated video.
Further, the building processing journey that processor 1001 can call the video user stored in memory 1003 to draw a portrait
Sequence also executes following operation:
Obtain the description text of each pre- setting video;
Obtain the corresponding multiple crucial term vectors of description text of the pre- setting video;
Obtain the weight of each crucial term vector;
The keyword of the pre- setting video is determined from each keyword according to the weight.
Further, the building processing journey that processor 1001 can call the video user stored in memory 1003 to draw a portrait
Sequence also executes following operation:
Obtain first hit-count of the crucial term vector in the description text of corresponding pre- setting video;
Obtain the second hit-count in the description text of pre- setting video of the crucial term vector in preset quantity;
The weight of the crucial term vector is determined according to first hit-count and second hit-count.
Further, the building processing journey that processor 1001 can call the video user stored in memory 1003 to draw a portrait
Sequence also executes following operation:
Judge whether the crucial term vector is located at the title or name entity of the description text of corresponding pre- setting video;
When the keyword vector is located at the title or name entity of the description text of corresponding pre- setting video, according to institute
The first modifying factor for stating weight generates revised weight;
When the crucial term vector is not located at the title or name entity of the description text of the pre- setting video, according to institute
The second modifying factor for stating weight generates revised weight, wherein first modifying factor is greater than second modifying factor
Son.
Further, the building processing journey that processor 1001 can call the video user stored in memory 1003 to draw a portrait
Sequence also executes following operation:
The keyword of the associated video is obtained according to the keyword of the pre- setting video and the associated video;
The keyword of each associated video is obtained in the corresponding keyword of video that the target user browsed
The first hit-count;
The keyword of each associated video is obtained in the corresponding keyword of video that the target user watched
The second hit-count;
Obtain predetermined registration operation video corresponding keyword of the keyword in the target user of each associated video
In third hit-count;
According to first hit-count, second hit-count and the third hit-count from the associated video
Keyword in determine the corresponding keyword of the historical operation information.
Further, the building processing journey that processor 1001 can call the video user stored in memory 1003 to draw a portrait
Sequence also executes following operation:
Obtain the historical operation video information of the user except the target user;
It is grasped according to the history of the user except the historical operation video information of the target user and the target user
The scoring of each associated video is obtained as video information;
The interest video of the target user is determined according to the scoring of each associated video;
It is that the target user constructs portrait according to the interest video and the keyword.
Further, the building processing journey that processor 1001 can call the video user stored in memory 1003 to draw a portrait
Sequence also executes following operation:
The video historical operation information of user is updated in real time or periodically.
Referring to Fig. 2, first embodiment of the invention provides a kind of construction method of video user portrait, which comprises
Step S10, historical operation video information and the historical operation video information for obtaining target user are corresponding
Keyword;
Target user in this step is the video user of portrait to be built, and the historical operation information of target user derives from
The user access information that the server for the website that user is accessed is stored.User access information includes the day that user accesses video
Will, user watch list of videos and/or user orders list of videos.Server stores the access information of all users, needs
It is extracted from the access information of all users according to the mark of target user and preset period of time related to target user
Historical operation video information.
When obtaining the historical operation video information of target user, target is obtained according to following steps S101~S102 and is used
The corresponding keyword of the historical operation information at family:
Step S101 determines the associated video of the target user according to the historical operation video information;
Available target user accessed from the historical operation video information of target user list of videos, target are used
The list of videos that the list of videos and target user that family was watched are ordered, carries out data point to the information in these list of videos
Analyse and handle the available target user may interested video, using these target users may interested video as
The associated video of target user.
It is corresponding to obtain the historical operation information according to the keyword of pre- setting video and the associated video by step S102
Keyword.
In the associated video that target user has been determined, according to the mark of associated video from the keyword of pre- setting video
With obtaining the keyword of associated video.For example, there is 100 pre- setting videos, each pre- setting video is corresponding with 10 keywords, then
The keyword of a total of 1000 pre- setting videos.The keyword of this 1000 pre- setting videos is with [default video identifier: keyword word
Group] data mode storage, wherein default video identifier can be video ID or video name, therefore can be regarded according to association
The mark of frequency is matched in the keyword of the pre- setting video of storage, the keyword of associated video is obtained, further according to preset
Rule filters out Partial key word from the keyword of associated video, believes Partial key word as the historical operation of target user
It ceases corresponding keyword and carries out subsequent processing.
It should be noted that default video includes all videos that the be supplied to user in website accesses, website each
Video all corresponds to one or more keywords.For example, the keyword of film " the wandering earth " has: science fiction, Liu Cixin, shock,
Light Jupiter, Wu Jing, milestone etc..
In the present embodiment, the corresponding keyword of pre- setting video carries out data mining by the description text to video, i.e.,
Data modeling is carried out to handle to obtain.The descriptive text of video includes the video essential information description of Website server itself storage
The text of the diversified forms such as news web page, blog, public platform on text, internet.Pass through the descriptive text to video
Data modeling processing, the keyword for extracting video are stored as the feature of video.
Associated video is obtained based on video access historical information of the target user within one period, more due to having
Access data sample for analyzing and handling, based on associated video obtain keyword as a kind of characteristic information, Ke Yicong
Generally relatively accurately portray the video preference of the target user.
Step S20 is that the target user constructs portrait according to the historical operation video information and the keyword.
Believed based on the general characteristic that can obtain reflection target user's video preference to the keyword obtained in last step
Breath, therefore the keyword can be used to construct the portrait of target user.Relative to the view obtained from keyword to target user
The good totality of frequency deviation is portrayed, it is also necessary to obtain additional characteristic information meticulously to portray the video preference of target user.
The historical operation video information of target user contains the target user to the detailed of a variety of operation behaviors of video
With complete information, such as the time rail that user's viewing, browsing, comment or the video name shared and these behaviors occur
Mark information can pass through the classification processing and modeling to video name and temporal information according to the needs of different application scenarios
Analysis, extracts the characteristic information that can meticulously portray user video behavior, utilizes this feature information and mesh obtained above
The overall video preference profiles information of mark user is that target user constructs portrait jointly.User video behavior can meticulously be portrayed
Characteristic information can be the interested video of target user, target user most likes to see the period of video or target user
Pay close attention to the behavioural habits etc. of video.
A kind of possible implementation method is that obtaining target user in the historical operation video information according to target user can
It is true using historical operation acquiring video information target user according to following step S201~S203 when the interested associated video of energy
Just interested interest video is that target user constructs portrait with the interest video and keyword jointly.
Step S201 obtains the historical operation video information of the user except the target user;
Step S202, according to the use except the historical operation video information of the target user and the target user
The historical operation video information at family obtains the scoring of each associated video;
Available user watches video, browsing video, collection video from the historical operation video information of target user
Historical operation video information with the comment much informations such as video, other users except target user also includes same type of
Much information can obtain target user to each associated video based on the comprehensive analysis processing of the much information to all users
Scoring.
Such as the associated video for each target user to be scored, first believe from the historical operation video of target user
Target user is obtained in breath to obtain to the viewing duration of the associated video, then from the historical operation information of other users to the pass
Join the viewing duration of video, can then be taken normalized based on the viewing duration of all users for watching the associated video
Method more accurately obtains the scoring of the viewing to the associated video.
For example, being directed to A video, the viewing duration of 10 users is respectively (1,2,3,4,5,6,7,8,9,10) hour, with
Be used as benchmark within longest viewing time 10 hours, obtain 10 users to the viewing of this video scoring for (0.1,0.2,0.3,
0.4,0.5,0.6,0.7,0.8,0.9,1).
It should be noted that associated video scoring obtained can be according to difference for each associated video to be scored
Preset rules obtained by the different behavior integration of target user.For example, preset rules are that target user occurs to browse video
Behavior obtains 0.1 point, and target user's collection video behavior occurs and obtains 0.2 point, and target user's comment video behavior occurs and obtains 0.3 point,
Target user's watching behavior score is obtained according to the viewing duration COMPREHENSIVE CALCULATING of all users, and final accumulative every score is closed
Join the final scoring of video.
Step S203 determines the interest video of the target user according to the scoring of each associated video;
Each associated video is ranked up according to the height of scoring, according to sequence from high to low, chooses preset quantity
Interest video of the associated video as target user.Such as preset quantity is 10, and scoring is come to the associated video of top ten list
Interest video as target user.
Step S204 is that the target user constructs portrait according to the interest video and the keyword.
After the interest video and keyword for obtaining target user, interest video and keyword are inputted into preset user
Model obtains the portrait of target user.
Such as in a kind of pre-set user model, the model U of target user ii: Ui={ Ii, RHij, Pij}。
IiRepresent the essential information of user i, Ii=gender=0, age=4 ... };
RHijInterest the video ID, j for representing user i take 1~20, RHi={ Ni1, Ni2..., Ni20};
PijRepresenting the keyword for the video behavioural characteristic that user i can be represented totally in i period of user, i takes 1~
20, Pij={ ki1, ki2..., ki20}。
It can be that target user carries out personalized push away based on the portrait of the target user after the portrait for obtaining target user
It recommends and is analyzed with further behavioural characteristic.
In the present embodiment, further, the video historical operation information that can in real time or periodically update user, can
With the video historical operation information based on update, more accurate user's portrait is constructed.
In the present embodiment, according to the historical operation video information of target user and the historical operation video information pair
The keyword answered is that target user constructs portrait, since the feature extracted from historical operation video information careful can portray mesh
User behavior feature is marked, while the corresponding keyword of historical operation video information reflects target user behavior spy on the whole
Point so that user's portrait of building not only has the characteristics of general outline, but also can be careful embodiment user the characteristics of, thus provide
A kind of construction method of more accurate user's portrait.
Further, referring to Fig. 3, second embodiment is proposed based on the first embodiment of the present invention, a kind of video use is provided
Family portrait construction method, the present embodiment the following steps are included:
Step S30 obtains the description text of each pre- setting video;
Default video includes all videos of the be supplied to user's access in website, and the descriptive text of pre- setting video includes net
The video essential information of site server itself storage describes a variety of shapes such as text, the news web page on internet, blog, public platform
The text of formula.By the way that the description text for passing to preset webpage capture script and obtaining the video being stored on internet will be identified
This.
Step S40 obtains the corresponding multiple crucial term vectors of the default video presentation text;
In this step, obtained using preparatory trained word vector the corresponding multiple words of description text of pre- setting video to
Amount can also filter out part from multiple term vectors by obtained multiple term vectors all one by one as crucial term vector
Term vector is as crucial term vector.
Step S50 obtains weight shared by each crucial term vector;
The crucial term vector quantity usually got from the description text of pre- setting video is more, in order to select most
The keyword that default video features can be represented needs to obtain the weight of each crucial term vector, last basis according to preset rules
The weight size of each key term vector therefrom determines the keyword of pre- setting video.
Specifically, it when obtaining the weight of each crucial term vector, can be carried out according to following step S501~S503: step
Rapid S501 obtains first hit-count of the crucial term vector in the description text of corresponding pre- setting video;Step S502 is obtained
Second hit-count of the crucial term vector in the description text of the pre- setting video of preset quantity;Step S503, according to the first life
Middle number and the second hit-count determine weight shared by each crucial term vector.Wherein, the pre- setting video of preset quantity
Can be all pre- setting videos, be also possible to according to the popular degree of pre- setting video select the pre- setting video in part, pass through by
Description text of the crucial term vector in corresponding pre- setting video is matched to obtain the first hit-count, passes through crucial term vector
It is matched to obtain the second hit-count in the default video presentation text of preset quantity.
In the present embodiment, the weight of each crucial term vector can be obtained according to following formula:
wjk=tfjk*idf
tfjk=count (SJk, j)/size(j)
In above-mentioned formula, wjkIndicate j-th of j-th of the pre- setting video weight for describing the corresponding k key term vector of text,
count(SIk, j) indicate frequency of occurrence of the key term vector k in the description text of j-th of pre- setting video;Size (j) is jth
The number of all keywords in a description text;NkIndicate the description text number containing k key term vector;N is indicated certain
In measurement period, the number of the description text of all pre- setting videos.
When obtaining the weight of each crucial term vector by above-mentioned formula, due to the description text of j-th of pre- setting video
It is obtained according to the associated description text of multiple j-th of video, k keyword is in the description text of corresponding j-th pre- setting video
Frequency of occurrence is higher, shows the characteristics of k keyword more can represent j-th of video, therefore the weight of k keyword is higher, meanwhile,
K keyword frequency of occurrence in other pre- setting videos is higher, illustrates that k keyword is more possible to represent this most of pre- setting video
Common feature, the conspicuousness relative to j-th of video is lower, therefore the weight of k keyword is lower.
Step S60 obtains the keyword of the pre- setting video according to the weight.
In the present embodiment, the power of each crucial term vector in the description text of pre- setting video is obtained by preset algorithm
It is worth, and therefrom selectes the keyword of pre- setting video according to the weight size of crucial term vector, constructing can the default view of accurate characterization
The keywords database of frequency feature is just able to accurately construct user's portrait based on this keywords database.
Further, referring to Fig. 4,3rd embodiment is proposed based on the second embodiment of the present invention, a kind of video use is provided
The construction method of family portrait, the present embodiment is after step S503 further include:
Step S70, judges whether the crucial term vector is located at the title or life of the description text of corresponding pre- setting video
Name entity;
The description text source of pre- setting video in video essential information describe text, the news web page on internet, blog,
The text of the diversified forms such as public platform, these texts have corresponding text subject, the corresponding pass of description text of pre- setting video
Keyword vector be obtained from the text of above-mentioned diversified forms can a series of vocabulary relevant to text subject.It is general and
Speech, when keyword vector is located at the title of the description text of default video, crucial term vector directly shows text subject at this time,
Or when keyword vector is located at the name entity of default video, show that crucial term vector is important at this time, therefore judge to close
Whether keyword vector is located at the title or name entity of the description text of corresponding pre- setting video, is further obtained according to judging result
Obtain the weight of more accurate keyword.
It should be noted that name entity is intrinsic title, abbreviation and other unique identifications in text, 7 are generally included
Kind classification: personage, mechanism, place, date, time, money and percentage.
Step S80, when the keyword vector is located at the title or name entity of the description text of corresponding pre- setting video
When, revised weight is generated according to the first modifying factor of the weight;
Step S90, when the crucial term vector is not located at the title or name entity of the description text of the pre- setting video
When, revised weight is generated according to the second modifying factor of the weight, wherein first modifying factor is greater than described the
Two modifying factors.
When by judging that discovery keyword vector is located at the title or name entity of the description text of corresponding pre- setting video
When, show that the topic relativity of the key term vector and text is larger, it should assign higher weight, pass through the first amendment at this time
The factor corrects weight;When the title or life of the description text by judging to find that crucial term vector is not located at corresponding pre- setting video
When name entity, show that the topic relativity of the key term vector and text is smaller, it should assign lower weight, pass through the at this time
One modifying factor corrects weight.
Specifically, revised weight can be obtained according to the result of judgement according to following formula, wherein when judgement is closed
When keyword vector is located at the title or name entity of the description text of corresponding pre- setting video, w > 1, on the contrary w < 1:
wjK=w*tfjk*idf
tfjk=count (SJk, j)/size(j)
In the present embodiment, by judging whether crucial term vector is located at the title of the description text of corresponding pre- setting video
Or name entity, the weight of more accurate keyword is further obtained according to judging result, constructing based on this can accurate table
The user's portrait for levying user characteristics.
Further, referring to Fig. 5, fourth embodiment is proposed based on the first embodiment of the present invention, a kind of video use is provided
The construction method of family portrait, the associated video include that the video that browsed of the target user, the target user watched
Video and the target user predetermined registration operation video, the present embodiment includes: in step s 102
Step S1021 obtains the pass of the associated video according to the keyword of the pre- setting video and the associated video
Keyword;
The pass of target user is being determined from all pre- setting videos according to the historical operation video information of target user
After joining video, the key of associated video can be indexed from the corresponding keywords database of pre- setting video according to the mark of associated video
Word.
Step S1022 obtains the keyword of each associated video in the video that the target user browsed
First hit-count;
Step S1023 obtains the keyword of each associated video in the video that the target user watched
Second hit-count;
Step S1024 obtains the keyword of each associated video in the predetermined registration operation video of the target user
Third hit-count;
Step S1025 is obtained according to first hit-count, second hit-count and the third hit-count
Total hit-count of the keyword of each associated video in the keyword of the associated video;
Step S1026 determines the historical operation according to total hit-count from the keyword of the associated video
The corresponding keyword of information.
Since the historical operation video information of target user contains the behavior of a variety of operation videos of target user, and it is different
The operation video behavior of type is different to the importance for the behavioral characteristic for indicating target user, therefore can be first by target user's
The corresponding associated video of historical operation information is classified according to behavior type, such as is classified as the video browsed, was watched
Video and predetermined registration operation video, then be based on sorted associated video, the keyword of statistical correlation video is in each classification
Frequency of occurrence in associated video keyword, the statistical result to keyword in different classes of associated video keyword assign not
Same weight, is weighted the keyword that operation obtains each associated video and occurs in all target user's historical operation behaviors
Total frequency of occurrence.
It should be noted that predetermined registration operation video includes that target user is shared with the video of other users, other users point
It enjoys to the video of target user, target user's collection or video of comment etc..
One mathematics modeling example of the above process is as follows:
Wherein, there is the number of keyword w in the historical behavior of Frequency (u, w) expression target user u;
frequency1Time that keyword w keyword occurs in the corresponding keyword of M video that (i, w) expression user u is browsed in the recent period
Number;frequency2(j, w) indicates that user had keyword w in the corresponding keyword of N number of video of watching behavior to occur in the recent period
Number;frequency3(k, w) indicates recent user by crucial in the keyword for the L video that other users are recommended or are shared
The number that word w occurs;Wherein α, beta, gamma are adjustable parameter, according to shadow of the operation behaviors all kinds of in application scenarios in user modeling
It rings and determines.
Finally total gone out according to what the keyword of each associated video occurred in all target user's historical operation behaviors
The height of occurrence number chooses historical operation information corresponding keyword of the keyword as target user of preset quantity.
In the present embodiment, by the corresponding associated video of historical operation information to target user according to behavior type into
Row classification, then frequency of occurrence of the keyword in the associated video keyword of each classification of statistical correlation video, to keyword
Different weights is assigned in the statistical result of different classes of associated video keyword, it is available more accurate to characterize use
The corresponding keyword of the historical operation information of family behavioural characteristic, to provide the accuracy of constructed user's portrait.
The present invention also provides a kind of server, which includes: memory, processor and is stored on the memory
And the portrait for the video user that can be run on the processor constructs processing routine, the portrait building processing of the video user
The step of portrait construction method of the video user is realized when program is executed by the processor.
In addition, the embodiment of the present invention also proposes a kind of computer readable storage medium, the computer readable storage medium
On be stored with video user portrait building processing routine, the video user portrait building processing routine be executed by processor
The step of portrait construction method of video user described in Shi Shixian.
It should be noted that, in this document, the terms "include", "comprise" or its any other variant are intended to non-row
His property includes, so that the process, method, article or the system that include a series of elements not only include those elements, and
And further include other elements that are not explicitly listed, or further include for this process, method, article or system institute it is intrinsic
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including being somebody's turn to do
There is also other identical elements in the process, method of element, article or system.
The serial number of the above embodiments of the invention is only for description, does not represent the advantages or disadvantages of the embodiments.
Through the above description of the embodiments, those skilled in the art can be understood that above-described embodiment side
Method can be realized by means of software and necessary general hardware platform, naturally it is also possible to by hardware, but in many cases
The former is more preferably embodiment.Based on this understanding, technical solution of the present invention substantially in other words does the prior art
The part contributed out can be embodied in the form of software products, which is stored in one as described above
In storage medium (such as ROM/RAM, magnetic disk, CD), including some instructions are used so that terminal device (it can be mobile phone,
Computer, server, air conditioner or network equipment etc.) execute method described in each embodiment of the present invention.
The above is only a preferred embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (10)
1. a kind of construction method of video user portrait, which is characterized in that the construction method of the video user portrait include with
Lower step:
Obtain the historical operation video information and the corresponding keyword of the historical operation video information of target user;
It is that the target user constructs portrait according to the historical operation video information and the keyword.
2. the construction method of video user portrait as described in claim 1, which is characterized in that described to obtain the historical operation
The step of video information corresponding keyword includes:
The associated video of the target user is determined according to the historical operation video information;
The corresponding keyword of the historical operation information is obtained according to the keyword of pre- setting video and the associated video.
3. the construction method of video user portrait as claimed in claim 2, which is characterized in that the structure of the video user portrait
Construction method is further comprising the steps of:
Obtain the description text of each pre- setting video;
Obtain the corresponding multiple crucial term vectors of description text of the pre- setting video;
Obtain the weight of each crucial term vector;
The keyword of the pre- setting video is determined from each keyword according to the weight.
4. the construction method of video user portrait as claimed in claim 3, which is characterized in that described to obtain each key
The step of weight of term vector includes:
Obtain first hit-count of the crucial term vector in the description text of corresponding pre- setting video;
Obtain the second hit-count in the description text of pre- setting video of the crucial term vector in preset quantity;
The weight of the crucial term vector is determined according to first hit-count and second hit-count.
5. the construction method of video user portrait as claimed in claim 4, which is characterized in that described according to first hit
Number and second hit-count determined after the step of weight of the crucial term vector further include:
Judge whether the crucial term vector is located at the title or name entity of the description text of corresponding pre- setting video;
When the keyword vector is located at the title or name entity of the description text of corresponding pre- setting video, according to the power
First modifying factor of value generates revised weight;
When the crucial term vector is not located at the title or name entity of the description text of the pre- setting video, according to the power
Second modifying factor of value generates revised weight, wherein first modifying factor is greater than second modifying factor.
6. the construction method of video user portrait as claimed in claim 2, which is characterized in that the associated video includes described
The predetermined registration operation video of video, the video that the target user watched and the target user that target user browsed, institute
State the step of corresponding keyword of the historical operation information is obtained according to the keyword and the associated video of pre- setting video packet
It includes:
The keyword of the associated video is obtained according to the keyword of the pre- setting video and the associated video;
Obtain of the keyword of each associated video in the corresponding keyword of video that the target user browsed
One hit-count;
Obtain of the keyword of each associated video in the corresponding keyword of video that the target user watched
Two hit-counts;
The keyword of each associated video is obtained in the corresponding keyword of predetermined registration operation video of the target user
Third hit-count;
According to first hit-count, second hit-count and the third hit-count from the pass of the associated video
The corresponding keyword of the historical operation information is determined in keyword.
7. the construction method of video user portrait as claimed in claim 2, which is characterized in that described according to the historical operation
Video information and the keyword are that the step of target user constructs portrait includes:
Obtain the historical operation video information of the user except the target user;
It is regarded according to the historical operation of the user except the historical operation video information of the target user and the target user
Frequency information obtains the scoring of each associated video;
The interest video of the target user is determined according to the scoring of each associated video;
It is that the target user constructs portrait according to the interest video and the keyword.
8. the construction method of video user portrait as described in claim 1, which is characterized in that the structure of the video user portrait
Construction method is further comprising the steps of:
The video historical operation information of user is updated in real time or periodically.
9. a kind of server, which is characterized in that the construction device of the video user portrait includes: memory, processor and deposits
The building control program for the video user portrait that can be run on the memory and on the processor is stored up, the video is used
The building control program of family portrait realizes that video described in any item of the claim 1 to 8 such as is used when being executed by the processor
The step of construction method of family portrait.
10. a kind of computer readable storage medium, which is characterized in that be stored with video use on the computer readable storage medium
The building of family portrait controls program, and the building control program of the video user portrait is realized when being executed by processor as right is wanted
The step of construction method of the portrait of video user described in asking any one of 1 to 8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910332502.7A CN110175264A (en) | 2019-04-23 | 2019-04-23 | Construction method, server and the computer readable storage medium of video user portrait |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910332502.7A CN110175264A (en) | 2019-04-23 | 2019-04-23 | Construction method, server and the computer readable storage medium of video user portrait |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110175264A true CN110175264A (en) | 2019-08-27 |
Family
ID=67690030
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910332502.7A Pending CN110175264A (en) | 2019-04-23 | 2019-04-23 | Construction method, server and the computer readable storage medium of video user portrait |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110175264A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111028572A (en) * | 2019-12-31 | 2020-04-17 | 浙江正元智慧科技股份有限公司 | Online education platform |
CN111047360A (en) * | 2019-12-16 | 2020-04-21 | 北京搜狐新媒体信息技术有限公司 | Data processing method and system based on visual portrait |
CN111079056A (en) * | 2019-10-11 | 2020-04-28 | 深圳壹账通智能科技有限公司 | Method, device, computer equipment and storage medium for extracting user portrait |
CN111368141A (en) * | 2020-03-18 | 2020-07-03 | 腾讯科技(深圳)有限公司 | Video tag expansion method and device, computer equipment and storage medium |
CN112818251A (en) * | 2021-04-13 | 2021-05-18 | 腾讯科技(深圳)有限公司 | Video recommendation method and device, electronic equipment and storage medium |
CN113627797A (en) * | 2021-08-12 | 2021-11-09 | 深圳平安智汇企业信息管理有限公司 | Image generation method and device for employee enrollment, computer equipment and storage medium |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140201038A1 (en) * | 2007-02-01 | 2014-07-17 | 7 Billion People, Inc. | Dynamic Reconfiguration of Web Pages Based on User Behavioral Portrait |
WO2016054908A1 (en) * | 2014-10-10 | 2016-04-14 | 中兴通讯股份有限公司 | Internet of things big data platform-based intelligent user profiling method and apparatus |
CN106294783A (en) * | 2016-08-12 | 2017-01-04 | 乐视控股(北京)有限公司 | A kind of video recommendation method and device |
CN106940705A (en) * | 2016-12-20 | 2017-07-11 | 上海掌门科技有限公司 | A kind of method and apparatus for being used to build user's portrait |
CN107124653A (en) * | 2017-05-16 | 2017-09-01 | 四川长虹电器股份有限公司 | The construction method of TV user portrait |
CN108694223A (en) * | 2018-03-26 | 2018-10-23 | 北京奇艺世纪科技有限公司 | The construction method and device in a kind of user's portrait library |
-
2019
- 2019-04-23 CN CN201910332502.7A patent/CN110175264A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140201038A1 (en) * | 2007-02-01 | 2014-07-17 | 7 Billion People, Inc. | Dynamic Reconfiguration of Web Pages Based on User Behavioral Portrait |
WO2016054908A1 (en) * | 2014-10-10 | 2016-04-14 | 中兴通讯股份有限公司 | Internet of things big data platform-based intelligent user profiling method and apparatus |
CN106294783A (en) * | 2016-08-12 | 2017-01-04 | 乐视控股(北京)有限公司 | A kind of video recommendation method and device |
CN106940705A (en) * | 2016-12-20 | 2017-07-11 | 上海掌门科技有限公司 | A kind of method and apparatus for being used to build user's portrait |
CN107124653A (en) * | 2017-05-16 | 2017-09-01 | 四川长虹电器股份有限公司 | The construction method of TV user portrait |
CN108694223A (en) * | 2018-03-26 | 2018-10-23 | 北京奇艺世纪科技有限公司 | The construction method and device in a kind of user's portrait library |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111079056A (en) * | 2019-10-11 | 2020-04-28 | 深圳壹账通智能科技有限公司 | Method, device, computer equipment and storage medium for extracting user portrait |
CN111047360A (en) * | 2019-12-16 | 2020-04-21 | 北京搜狐新媒体信息技术有限公司 | Data processing method and system based on visual portrait |
CN111047360B (en) * | 2019-12-16 | 2024-04-09 | 北京搜狐新媒体信息技术有限公司 | Data processing method and system based on visual portraits |
CN111028572A (en) * | 2019-12-31 | 2020-04-17 | 浙江正元智慧科技股份有限公司 | Online education platform |
CN111368141A (en) * | 2020-03-18 | 2020-07-03 | 腾讯科技(深圳)有限公司 | Video tag expansion method and device, computer equipment and storage medium |
CN112818251A (en) * | 2021-04-13 | 2021-05-18 | 腾讯科技(深圳)有限公司 | Video recommendation method and device, electronic equipment and storage medium |
CN113627797A (en) * | 2021-08-12 | 2021-11-09 | 深圳平安智汇企业信息管理有限公司 | Image generation method and device for employee enrollment, computer equipment and storage medium |
CN113627797B (en) * | 2021-08-12 | 2023-11-14 | 深圳平安智汇企业信息管理有限公司 | Method, device, computer equipment and storage medium for generating staff member portrait |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10824682B2 (en) | Enhanced online user-interaction tracking and document rendition | |
US10671620B2 (en) | Method for recommending a teacher in a network teaching system | |
CN110175264A (en) | Construction method, server and the computer readable storage medium of video user portrait | |
US20210209156A1 (en) | System and method for providing content to users based on interactions by similar other users | |
TWI636416B (en) | Method and system for multi-phase ranking for content personalization | |
JP5731250B2 (en) | System and method for recommending interesting content in an information stream | |
CN110209827B (en) | Search method, search device, computer-readable storage medium, and computer device | |
CN107862022B (en) | Culture resource recommendation system | |
US20110196927A1 (en) | Social Networking Application Using Posts to Determine Compatibility | |
CN110532451A (en) | Search method and device for policy text, storage medium, electronic device | |
US20160055541A1 (en) | Personalized recommendation system and methods using automatic identification of user preferences | |
US20160071117A1 (en) | System and method for using marketing automation activity data for lead prioritization and marketing campaign optimization | |
US10402479B2 (en) | Method, server, browser, and system for recommending text information | |
US20150332373A1 (en) | Method and system for pushing mobile application | |
WO2011080899A1 (en) | Information recommendation method | |
CN109889891B (en) | Method, device and storage medium for acquiring target media file | |
GB2507667A (en) | Targeted advertising based on momentum of activities | |
WO2014130843A1 (en) | System and method for revealing correlations between data streams | |
KR20070058615A (en) | Modifying an on-line dating search using inline editing | |
CN110134845A (en) | Project public sentiment monitoring method, device, computer equipment and storage medium | |
CN108509499A (en) | A kind of searching method and device, electronic equipment | |
CN103577504A (en) | Method and device for putting personalized contents | |
CN111651666A (en) | User theme recommendation method and device, computer equipment and storage medium | |
CN113869931A (en) | Advertisement putting strategy determining method and device, computer equipment and storage medium | |
JP5228584B2 (en) | Interest information identification system, interest information identification method, and interest information identification program |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190827 |