CN112488768A - Feature extraction method, feature extraction device, storage medium, and electronic apparatus - Google Patents

Feature extraction method, feature extraction device, storage medium, and electronic apparatus Download PDF

Info

Publication number
CN112488768A
CN112488768A CN202011457798.4A CN202011457798A CN112488768A CN 112488768 A CN112488768 A CN 112488768A CN 202011457798 A CN202011457798 A CN 202011457798A CN 112488768 A CN112488768 A CN 112488768A
Authority
CN
China
Prior art keywords
information
vertex
label
data
demander
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
Application number
CN202011457798.4A
Other languages
Chinese (zh)
Inventor
黄崇远
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.)
Guangdong Oppo Mobile Telecommunications Corp Ltd
Shenzhen Huantai Technology Co Ltd
Original Assignee
Guangdong Oppo Mobile Telecommunications Corp Ltd
Shenzhen Huantai Technology 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 Guangdong Oppo Mobile Telecommunications Corp Ltd, Shenzhen Huantai Technology Co Ltd filed Critical Guangdong Oppo Mobile Telecommunications Corp Ltd
Priority to CN202011457798.4A priority Critical patent/CN112488768A/en
Publication of CN112488768A publication Critical patent/CN112488768A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • 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
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241—Advertisements
    • G06Q30/0251—Targeted advertisements
    • G06Q30/0269—Targeted advertisements based on user profile or attribute
    • G06Q30/0271—Personalized advertisement
    • 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
    • G06F16/906—Clustering; Classification
    • 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
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06315—Needs-based resource requirements planning or analysis
    • 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
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241—Advertisements
    • G06Q30/0242—Determining effectiveness of advertisements
    • 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
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241—Advertisements
    • G06Q30/0277—Online advertisement

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Strategic Management (AREA)
  • Development Economics (AREA)
  • Finance (AREA)
  • Accounting & Taxation (AREA)
  • Theoretical Computer Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Human Resources & Organizations (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Databases & Information Systems (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Educational Administration (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The disclosure provides a feature extraction method, a feature extraction device, a storage medium and electronic equipment, and relates to the technical field of data processing. The feature extraction method comprises the following steps: acquiring association relations between a plurality of demanders and a plurality of information labels from information delivery data; establishing an information release relation graph based on the incidence relation, wherein the vertex of the information release relation graph comprises the demand party and the information label; forming a plurality of vertex sequences through paths in the information delivery relation graph; and extracting the characteristic data of at least one vertex according to the vertex sequence. The accuracy of extraction of the characteristics of the demander or the information label can be improved, and accurate matching of the demander and the information label is facilitated.

Description

Feature extraction method, feature extraction device, storage medium, and electronic apparatus
Technical Field
The present disclosure relates to the field of data processing technologies, and in particular, to a feature extraction method, a feature extraction device, a computer-readable storage medium, and an electronic device.
Background
With the popularization of the internet and the mobile internet, information delivery using the internet and the mobile internet as carriers has become a mainstream propaganda and popularization method. For example, information such as advertisements, news, and self-media articles is delivered to App (Application) pages and web pages, or is pushed to users in the form of short messages, desktop popups, App messages, and the like.
To improve the accuracy of information delivery, information tags are often used for targeted delivery, such as delivering advertisements from a sporting goods web owner to users with "sporting" tags. However, the related art often fails to accurately match a proper information tag for a demand side, so that information is released to a user group not interested in, and the releasing effect is affected.
Disclosure of Invention
The present disclosure provides a feature extraction method, a feature extraction device, a computer-readable storage medium, and an electronic device, thereby solving, at least to some extent, the problem of matching between a requesting party and an information tag.
Additional features and advantages of the disclosure will be set forth in the detailed description which follows, or in part will be obvious from the description, or may be learned by practice of the disclosure.
According to a first aspect of the present disclosure, there is provided a feature extraction method including: acquiring association relations between a plurality of demanders and a plurality of information labels from information delivery data; establishing an information release relation graph based on the incidence relation, wherein the vertex of the information release relation graph comprises the demand party and the information label; forming a plurality of vertex sequences through paths in the information delivery relation graph; and extracting the characteristic data of at least one vertex according to the vertex sequence.
According to a second aspect of the present disclosure, there is provided a feature extraction device including: the incidence relation acquisition module is configured to acquire incidence relations between a plurality of demanders and a plurality of information labels from the information delivery data; the relation graph establishing module is configured to establish an information release relation graph based on the incidence relation, and the vertex of the information release relation graph comprises the demand party and the information label; the vertex sequence extraction module is configured to form a plurality of vertex sequences through paths in the information delivery relation graph; a feature data extraction module configured to extract feature data of at least one vertex from the sequence of vertices.
According to a third aspect of the present disclosure, there is provided a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the feature extraction method of the first aspect described above and possible implementations thereof.
According to a fourth aspect of the present disclosure, there is provided an electronic device comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to perform the feature extraction method of the first aspect described above and possible implementations thereof via execution of the executable instructions.
The technical scheme of the disclosure has the following beneficial effects:
and establishing an information release relation graph based on the incidence relation between the demand party and the information label in the information release data, extracting a vertex sequence from the information release relation graph, and further extracting characteristic data according to the vertex sequence. On the one hand, the scheme for extracting the characteristics of the demander or the information tag is provided, the depth correlation among different vertexes is mined through a vertex sequence in the information delivery relation graph so as to obtain the densified characteristic data, the abstract characteristics of the demander or the information tag can be reflected, and the accuracy is high. On the other hand, the scheme can be realized based on historical record data of information delivery, and can be deployed in an off-line stage, so that the data processing amount in on-line practical application is reduced, and the processing efficiency of information delivery is improved.
Furthermore, the relevance between the demander and the information tag can be determined by utilizing the characteristic data obtained by the scheme, so that the matching problem of the demander and the information tag is solved, and the accurate positioning and the effective expansion of the information delivery user group of the demander are facilitated. The method is not limited to a fixed group any more, meanwhile, the situation that labels are used improperly possibly caused by artificial selection of information labels is avoided, the global optimal release recommendation scheme is favorably realized, and an actual release test is not needed when the information labels are recommended, so that the trial and error cost is reduced.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and together with the description, serve to explain the principles of the disclosure. It is apparent that the drawings in the following description are only some embodiments of the present disclosure, and that other drawings can be obtained from those drawings without inventive effort for a person skilled in the art.
FIG. 1 shows a schematic diagram of a system architecture in the present exemplary embodiment;
fig. 2 shows a flowchart of a feature extraction method in the present exemplary embodiment;
fig. 3 shows an information delivery relationship diagram in the present exemplary embodiment;
fig. 4 shows another information delivery relationship diagram in the present exemplary embodiment;
FIG. 5 illustrates a flow chart for extracting vertex sequences in the exemplary embodiment;
FIG. 6 shows a schematic diagram of an embedding model in the present exemplary embodiment;
FIG. 7 shows a flow chart for extracting feature data in the present exemplary embodiment;
fig. 8 shows a flowchart of another feature extraction method in the present exemplary embodiment;
fig. 9 is a diagram showing the structure of a feature extraction device in the present exemplary embodiment;
fig. 10 shows a block diagram of an electronic device in the present exemplary embodiment.
Detailed Description
Exemplary embodiments will now be described more fully with reference to the accompanying drawings. The exemplary embodiments, however, may be embodied in many different forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of embodiments of the disclosure. One skilled in the relevant art will recognize, however, that the subject matter of the present disclosure can be practiced without one or more of the specific details, or with other methods, components, devices, steps, and the like. In other instances, well-known technical solutions have not been shown or described in detail to avoid obscuring aspects of the present disclosure.
Furthermore, the drawings are merely schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus their repetitive description will be omitted. Some of the block diagrams shown in the figures are functional entities and do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different networks and/or processor devices and/or microcontroller devices.
The flow charts shown in the drawings are merely illustrative and do not necessarily include all of the steps. For example, some steps may be decomposed, and some steps may be combined or partially combined, so that the actual execution sequence may be changed according to the actual situation.
In one scheme of the related art, a method for expanding advertisement delivery is provided: firstly, historical related advertisements of currently delivered advertisements are obtained, wherein the historical related advertisements comprise the advertisements of the same type and the historical advertisements delivered by the same advertiser; then, analyzing the use condition of the related advertisement targeting, and generally judging the effect of the targeting condition by using indexes such as conversion rate and the like; then screening out an orientation condition with good effect as the extension orientation of the existing orientation condition; and finally, putting. However, in the scheme, the way of acquiring the related advertisements is limited, and if the number of the advertisements of the same type and the number of the advertisements of the same advertiser are small, the expanded targeting conditions are limited and the accuracy is low; moreover, the judgment of the effect of the orientation condition is more comprehensive, which can cause the screened orientation condition to be deviated from the actual situation (such as over high bid).
In another scheme of the related art, a method for evaluating the effect of two-round delivery is provided: firstly, manually combining labels, performing a first round of information release, and recovering released effect data; then, the manual combination labels of the first round are recombined, and information of the second round is released; and comparing the releasing effects of the first round and the second round, and selecting the label combination with good effect as the final label combination. However, in the scheme, the combined label of the first wheel is determined manually, so that the method has limitations, even if the combined label of the second wheel is optimized, the method cannot be separated from the basis of the combined label of the first wheel, and a global optimal scheme is difficult to obtain; in addition, the effect evaluation is carried out through the actual delivery, the prediction cannot be carried out in advance, and the effect cannot be accurately delivered to interested groups in the first delivery and the second delivery, so that the trial and error cost is increased.
In view of one or more of the above problems, exemplary embodiments of the present disclosure provide a feature extraction method, application scenarios of which include but are not limited to: and (4) extracting the characteristics of the demander or the information label for information delivery so as to more accurately match the demander and the information label and further realize the accurate positioning and effective expansion of the information delivery user group.
Fig. 1 shows a system architecture diagram of a feature extraction method operating environment. As shown in fig. 1, information delivery system 100 includes a demander party 110, a platform party 120, and a user 130. The demander 110 refers to a party with production information and delivery requirements, including advertisers, self-media, online stores, etc., and accesses the platform side 120 through a personal computer, a smart phone, a server, etc. The platform side 120 is a side providing an information delivery platform, and is generally an internet service provider, and is deployed with a relevant server and a database. The user 130 is an end user who logs in the platform through a client, a browser, or the like, and can view the delivered information on the platform.
The feature extraction method in the present exemplary embodiment may be performed by the platform side 120 and, in some cases, may also be performed by the demand side 110. Fig. 2 shows an exemplary flow of a feature extraction method, which may include:
step S210, acquiring the incidence relation between a plurality of demanders and a plurality of information labels from information delivery data;
step S220, establishing an information release relation graph based on the incidence relation, wherein the top point of the information release relation graph comprises a demand party and an information label;
step S230, forming a plurality of vertex sequences through paths in the information delivery relation graph;
step S240, extracting the characteristic data of at least one vertex according to the vertex sequence.
By the method, the information delivery relation graph is established based on the incidence relation between the demand side and the information label in the information delivery data, the vertex sequence is extracted from the information delivery relation graph, and the characteristic data is further extracted according to the vertex sequence. On the one hand, the scheme for extracting the characteristics of the demander or the information tag is provided, the depth correlation among different vertexes is mined through a vertex sequence in the information delivery relation graph so as to obtain the densified characteristic data, the abstract characteristics of the demander or the information tag can be reflected, and the accuracy is high. On the other hand, the scheme can be realized based on historical record data of information delivery, and can be deployed in an off-line stage, so that the data processing amount in on-line practical application is reduced, and the processing efficiency of information delivery is improved. Furthermore, the relevance between the demander and the information tag can be determined by utilizing the characteristic data obtained by the scheme, so that the matching problem of the demander and the information tag is solved, and the accurate positioning and the effective expansion of the information delivery user group of the demander are facilitated. The method is not limited to a fixed group any more, meanwhile, the situation that labels are used improperly possibly caused by artificial selection of information labels is avoided, the global optimal release recommendation scheme is favorably realized, and an actual release test is not needed when the information labels are recommended, so that the trial and error cost is reduced.
Each step in fig. 2 will be described in detail below.
In step S210, the association relationship between the plurality of demanders and the plurality of information tags is obtained from the information delivery data.
The information delivery data refers to historical data of information delivery. When information is released, an information label is generally marked on the information of a demand party, and the information is released to a user group corresponding to the information label. Therefore, each piece of information delivery data can be arranged into a binary array (demander and information label), and the demander and the information label in the binary array have an association relationship.
When the information delivery data is acquired, appropriate screening can be performed, so that the association relationship between the demander and the information tag in the information delivery data is more effective. Including but not limited to:
and screening according to the time range, so that the information delivery data in the latest preset time range (such as the latest month, the latest half year and the like) can be screened out, and the association relationship between the demand party and the information label is obtained. The business content and the scope of the demand side can be changed, and the obtained association relationship reflects the latest situation of the demand side.
And screening according to the release feedback data, so that the information release data with the release feedback data reaching a preset standard can be screened out, and the incidence relation between the demand party and the information label is obtained. The delivery feedback data is index data reflecting the delivery effect, for example, after the delivery information, index data such as click rate and conversion rate are counted and recorded in the information delivery data, that is, the delivery feedback data. The preset standard can be determined according to experience and actual requirements, the information releasing data with good releasing effect can be screened out through the preset standard, the relevance between a demand party and the information label is stronger, and the obtained relevance relation is more effective.
For example, the advertisement placement data of the last month is obtained, and each piece of data includes: advertisement, affiliated advertiser (i.e., demander), advertisement tag (i.e., information tag), CTR (Click Through Rate). Examples are as follows:
ad a, advertiser a, ad tag: basketball | football | European and American | novel, CTR _ a;
advertisement b, advertiser a, advertisement tag: novel | financing | elementary school education, CTR _ b;
ad c, advertiser B, ad tag: stock | financing | novel | high school education, CTR _ c;
ad d, advertiser C, ad tag: football primary education | korea drama, CTR _ d.
Screening out information delivery data with CTR being more than or equal to 0.5 (namely a preset standard), and if CTR _ a is 0.7, namely the information delivery data of the advertisement a meets the preset standard, acquiring the association relationship between an advertiser and an advertisement tag from the information delivery data, wherein the method comprises the following steps:
(advertiser A, basketball)
(advertiser A, football)
(advertiser A, European comedy)
(advertiser A, novel)
And processing other information delivery data meeting the preset standard in the same way to obtain the association relation between all advertisers and the advertisement tags.
In one embodiment, when step S210 is executed, the association relationship between the information tag and the tag category may also be obtained. Generally, information tags may be classified in advance to obtain a plurality of tag categories. A plurality of information tags are usually included in one tag category, and one information tag may belong to a plurality of tag categories. Multiple levels of label categories may also be set, for example, a plurality of secondary label categories are included under the primary label category, and a plurality of information labels are included under each secondary label category. Examples of information tag to tag class associations are provided below:
(basketball, sports)
(football, sports)
(European American, movie & TV play/leisure)
(novel, leisure)
(financing, finance)
(education, education in primary school)
In one embodiment, the association relationship between the demander and the information tag and the association relationship between the information tag and the tag category are combined, so as to obtain the association relationship among the demander, the information tag, and the tag category, for example, the following data format is obtained:
(advertiser A, basketball (sports), football (sports), European and American drama (movie, leisure), novel (leisure))
(advertiser A, novel (leisure), financing (finance), primary education (education))
(advertiser B, stock (finance), financing (finance), novel (leisure), high school education (education))
(advertiser C, football (sports), primary education (education), Korean play (movie, leisure))
With reference to fig. 2, in step S220, an information delivery relationship graph is established based on the association relationship, and a vertex of the information delivery relationship graph includes the demander and the information label.
And forming a vertex for each demand party and each information label respectively, and connecting the demand party with the incidence relation and the information labels to form an edge so as to obtain an information delivery relation graph. FIG. 3 illustrates an information placement relationship diagram formed by advertiser A, advertiser C, and their associated advertisement tags. The advertiser A has an association relationship with basketball, football, European and American dramas and novels, edges are arranged between vertexes of the advertiser A, and the advertiser C has an association relationship with football, Korean dramas and primary school education, and edges are arranged between vertexes of the advertiser C. All the incidence relations can be recorded through the information delivery relation graph.
In one embodiment, if the association relationship between the demand party and the information tag and the association relationship between the information tag and the tag category are obtained, an information delivery relationship graph can be established based on the two association relationships. The vertex of the information release relation graph comprises a demand party, an information label and a label category. Fig. 4 shows a graph of information delivery relationships collectively formed by advertisers, advertisement tags, and tag categories. Compared with fig. 3, the vertex of the label category is added, and the advertisement label with the association relationship has an edge with the label category.
The edges in the information delivery relationship graph may be unweighted, and all the edges may be considered to be equally weighted. Different weights may also be set for different edges. In an embodiment, an edge weight between the demander and the information tag may be set according to the delivery feedback data, for example, a click rate setting weight is adopted, a click rate CTR (advertiser a- > basketball) obtained when the advertisement of the advertiser a is delivered to the user group corresponding to the advertisement tag "basketball" in the last month is counted as an edge weight between two vertexes of the advertiser a and the advertisement tag "basketball", a CTR value may be normalized as an edge weight, and a CTR value below a certain value (e.g., 0.5) may be set so that no edge exists between the two vertexes.
With continued reference to fig. 2, in step S230, a plurality of vertex sequences are formed through the paths in the information delivery relationship diagram.
One or more paths may be extracted based on the connection relationships in the information delivery relationship graph. The paths are subgraphs in the graph formed by a plurality of sequentially connected vertexes, an edge exists between two adjacent vertexes in each path, and no repeated vertex exists in each path. The vertices in the path are sequentially extracted to form one or more vertex sequences. For example in fig. 3, advertiser a- > football- > advertiser C is a path, a vertex sequence (advertiser a, football, advertiser C) may be formed.
In one embodiment, the information delivery relationship graph may be walked according to a preset rule to obtain the path. The preset rule may include any one or more of the following:
firstly, if the current vertex is a demand side, the next vertex is an information label;
if the current vertex is an information label, the next vertex is a demand party;
thirdly, if the previous vertex is a demand party and the current vertex is an information label, the next vertex is a label category;
if the previous vertex is of the label type and the current vertex is of the information label, the next vertex is of the demand party;
the initial vertex of the path is a demand side;
sixthly, the ending peak of the path is a demand side;
seventhly, except for a starting vertex and an ending vertex, other vertexes in the path are all information labels or label types;
the number of the vertexes in the path is in a preset range, and the length of the path is actually limited.
In practical application, the rules can be combined and used at will according to requirements. For example:
sixthly, actually setting a circulation migration rule (marked as a circulation migration rule I) of a demand party- > an information label- > …;
and (4) actually setting a circulation wandering rule (marked as a circulation wandering rule two) of a demand party- > information label- > label category- > information label- > demand party- > ….
Under the above rules, random walk may be adopted in the actual walk, and the walk probabilities of different edges may also be set according to the edge weights. E.g. to vertex viThen, go through and v in the next stepiAdjacent vertexes v connected and satisfying a predetermined rulejThe probability of (c) is as follows:
Figure BDA0002829761970000091
wherein G represents an information delivery relation graph; n + (v)i) Is at the vertex viA set of connected adjacent vertices; c (v)i) Is and satisfies the vertex v of the predetermined ruleiThe next set of vertices; eijRepresenting a vertex viAnd vertex vjEdge between, MijRepresenting the weight of the edge.
In fig. 4, the first and second cyclic walking rules are respectively adopted to walk, and the following paths can be obtained:
the first route is advertiser A- > basketball- > advertiser D- > stock- > advertiser B- > financing.
Path two, advertiser a- > basketball- > sports- > football- > advertiser C- > primary school education- > high school education- > advertiser b.
It should be noted that different preset rules may be adopted, and different vertices are used as the starting vertex or the ending vertex, and paths with different lengths are set, so that a large number of paths meeting the requirements may be extracted from one information delivery relationship diagram. For each path, all vertices may be selected, or a part of vertices may be selected to form a vertex sequence, so that one path may correspondingly form a plurality of vertex sequences. Whereby a plurality of vertex sequences may be obtained.
In one embodiment, step S230 may include:
and sequentially extracting homogeneous vertexes from the path to obtain a vertex sequence.
The vertexes of the same type are homogeneous vertexes, for example, the advertiser A and the advertiser B are vertexes of a demand party and homogeneous vertexes, basketball and football are vertexes of information labels and homogeneous vertexes, and sports and leisure are vertexes of label categories and homogeneous vertexes; otherwise, the vertex is heterogeneous, such as the vertex of advertiser A and basketball.
The paths comprise different types of vertexes, and homogeneous vertexes are extracted when the vertexes are extracted, so that a vertex sequence consisting of the vertexes of the same type is obtained. In one embodiment, as shown in fig. 5, the sequentially extracting homogenous vertices from the path to obtain a vertex sequence may include:
step S510, extracting vertexes corresponding to the demand side from the path in sequence to obtain a demand side sequence;
step S520, sequentially extracting vertices corresponding to the information labels from the path to obtain an information label sequence.
For example, the vertex corresponding to the demander is extracted from the path one and the path two to obtain a demander sequence:
advertiser a, advertiser D, advertiser B.
Advertiser a, advertiser C, advertiser B.
Extracting vertexes corresponding to the information labels from the first path and the second path to obtain an information label sequence:
basketball, stock, financing.
Basketball, football, primary school education, high school education.
With continued reference to fig. 2, in step S240, feature data of at least one vertex is extracted according to the vertex sequence.
By establishing the information delivery relation graph and extracting the vertex sequence, the relation between the vertexes without direct association is established, and the relation between the vertexes is further mined. Such a relationship is embodied in a positional relationship between different vertices in the vertex sequence. Thereby feature data of the vertices can be extracted.
In one embodiment, step S240 may include:
and processing the vertex sequence by using the embedded model to obtain the characteristic data of at least one vertex.
Each vertex sequence is regarded as a sentence or a section of text, and the vertices are different words, so that the characteristics of the words can be embedded into a data space and processed in a word embedding mode to obtain the characteristic data of the vertices. The description will be given by taking an implementation mode CBOW (Continuous Bag of Words) of the embedded model word2vec as an example. The principle of CBOW can be seen in FIG. 6, where a vertex in the vertex sequence is input (in w)tRepresentation) of a context (i.e., w)tFront and rear vertices in a certain range, e.g. wt-c，…，wt-1，wt+1，…，wt+c) Predicting wtIt can be expressed as a conditional probability as follows:
P(wt|wt-c:wt+c) (2)
thus, the goal of the embedded model is to maximize the log-likelihood function of the conditional probability:
Figure BDA0002829761970000111
where T represents the length of the vertex sequence, i.e., the number of vertices in the sequence. The conditional probability can be obtained by a softmax function (normalized exponential function):
Figure BDA0002829761970000112
Figure BDA0002829761970000113
the vertex sequence obtained in step S230 is introduced into the embedded model, and the embedded model can be trained. And outputting feature data of different vertexes by using the middle layer of the trained embedded model, wherein the feature data can be dense Embedding vectors.
In one embodiment, referring to fig. 7, step S240 may include:
step S710, obtaining characteristic data of at least one demand side according to the demand side sequence;
and S720, obtaining the characteristic data of at least one information label according to the information label sequence.
Generally, a sequence of the demander and a sequence of the information tag can form one data set, and the embedded model can be trained by using the two data sets respectively, so as to output characteristic data of the demander and characteristic data of the information tag respectively.
For example, the corresponding feature data may be output to the advertiser and advertisement tag in fig. 3 or fig. 4, respectively:
the advertiser A: 0.3345, 0.1240, 0.1763, 0.6421, 0.0144, 0.3721, 0.1983, 0.6311;
the advertiser B: 0.5521, 0.1392, 0.1732, 0.3124, 0.0211, 0.1359, 0.2345, 0.5367;
and the advertiser C: 0.3340, 0.2321, 0.6161, 0.4223, 0.8315, 0.1482, 0.3984, 0.4124;
....
basketball: 0.0342, 0.1230, 0.7762, 0.6481, 0.6144, 0.1721, 0.8983, 0.2319;
football: 0.5223, 0.2392, 0.5731, 0.2124, 0.3211, 0.7359, 0.2315, 0.1362;
financing: 0.5341,0.2323,0.1169,0.4233,0.8317,0.2482,0.1981,0.6112.
In one embodiment, after obtaining the characteristic data of the requesting party and the characteristic data of the information tag, the following steps may be further performed:
and determining the correlation degree of the demander and the information label according to the characteristic data of the demander and the characteristic data of the information label.
For example, the feature data of the demand side and the feature data of the information tag are respectively used as different arrays, and the similarity between the two arrays is calculated. Or, the feature data may be in the form of a feature vector, and an inner product (or cosine similarity, euclidean distance, etc.) of the feature vector of the demand party and the feature vector of the information tag is calculated to obtain the correlation between the demand party and the information tag.
In one embodiment, any demander can be used as a target demander, and the correlation degree of the target demander and each information tag is determined so as to recommend the information tag for information delivery to the target demander. For example, the relevance of advertiser A and each advertisement label is calculated, and N advertisement labels with the highest relevance are determined to be recommended to advertiser A. Or rank the relevance of the ad tags for each advertiser as follows:
the advertiser A: european and American dailies, football, basketball, lottery, novel.
The advertiser B: live broadcast, short video, novel, basketball, loan.
And the advertiser C: korea drama, direct broadcast, novel, basketball, and football.
Therefore, the information label with higher relevance can be determined for the demand side, and the accurate positioning and the effective expansion of the user group in the information delivery process are facilitated.
In one embodiment, the correlation calculated from the feature data may be normalized, and the result may be referred to as a first correlation. And determining a second degree of correlation between the demander and the information tag according to the delivery feedback data, for example, counting click rates obtained when the advertisement of the advertiser A is delivered to user groups corresponding to different advertisement tags in the last month, and recording the normalized click rates as the second degree of correlation. The first degree of correlation is then weighted with the second degree of correlation:
final correlation w1 first correlation + w2 second correlation (6)
Where w1 and w2 are the weights of the first correlation and the second correlation, respectively, and can be empirically set, for example, w1 is 0.7 and w2 is 0.3. Therefore, the information of the characteristic data and the feedback data of the historical release are fused, and the obtained final correlation degree has higher accuracy.
In one embodiment, the relevancy of each demander and each information tag can be calculated by the method, and a relevancy data table is established so as to facilitate subsequent searching and use.
In one embodiment, an exemplary flow of the feature extraction method may be as shown in fig. 8, including:
step S801, acquiring information release data of the latest month;
s802, screening out information delivery data with higher CTR (if CTR is more than or equal to 0.5); then steps S803 to S808 are performed on the one hand and steps S809 and S810 are performed on the other hand;
step S803, determining the association relation among the demand party, the information labels and the label categories from the information delivery data;
step S804, establishing an information delivery relation graph based on the incidence relation;
step S805, according to a preset rule, performing random walk in an information delivery relation graph to obtain a plurality of paths;
step S806, extracting a demander sequence and an information tag sequence from the path;
step S807, processing the demander sequence and the information tag sequence by using the embedded model respectively to obtain the characteristic vectors of the demander and the information tag;
step S808, calculating the inner product of the characteristic vectors of the demand side and the information label to obtain a first correlation degree of the demand side and the information label;
step S809, calculating average CTR of the demander and the information label from the information release data;
step S810, carrying out normalization calculation through the statistical average CTR to obtain a second degree of correlation between the demand side and the information label;
step S811, the first correlation and the second correlation are weighted and calculated to obtain the final correlation between the demander and the information label.
Therefore, one or more information labels with the highest final relevance can be recommended for the demanders, and the method is favorable for determining a proper information delivery user group.
Exemplary embodiments of the present disclosure also provide a feature extraction apparatus. Referring to fig. 9, the feature extraction apparatus 900 may include:
an association relation obtaining module 910, configured to obtain association relations between a plurality of demanders and a plurality of information tags from information delivery data;
a relation graph establishing module 920 configured to establish an information delivery relation graph based on the association relation, where a vertex of the information delivery relation graph includes a demand party and an information label;
a vertex sequence extraction module 930 configured to form a plurality of vertex sequences through paths in the information delivery relationship graph;
a feature data extraction module 940 configured to extract feature data of at least one vertex from the sequence of vertices.
In one embodiment, the association obtaining module 910 is configured to:
and acquiring the association relation between the information tag and the tag category.
A relationship graph establishing module 920 configured to:
and establishing an information release relation graph based on the incidence relation between the demand party and the information label and the incidence relation between the information label and the label category, wherein the vertex of the information release relation graph comprises the demand party, the information label and the label category.
In one embodiment, the vertex sequence extraction module 930 is configured to:
and traveling in the information delivery relation graph according to a preset rule to acquire the path.
Wherein the preset rule comprises at least one of:
if the current vertex is the demand side, the next vertex is an information label;
if the current vertex is an information label, the next vertex is a demand side;
if the previous vertex is a demand side and the current vertex is an information label, the next vertex is a label type;
if the previous vertex is of the label type and the current vertex is of the information label, the next vertex is a demand party;
the initial vertex of the path is a demand side;
the end vertex of the path is a demand side;
except for a starting vertex and an ending vertex, other vertexes in the path are all information labels or label categories;
the number of vertices in the path is within a predetermined range.
In one embodiment, the vertex sequence extraction module 930 is configured to:
and sequentially extracting homogeneous vertexes from the path to obtain a vertex sequence.
In one embodiment, the vertex sequence extraction module 930 is configured to:
sequentially extracting vertexes corresponding to the demand side from the path to obtain a demand side sequence;
and sequentially extracting vertexes corresponding to the information labels from the path to obtain an information label sequence.
In one embodiment, the feature data extraction module 940 is configured to:
obtaining characteristic data of at least one demand side according to the demand side sequence;
and obtaining the characteristic data of at least one information label according to the information label sequence.
In one embodiment, the feature extraction apparatus 900 further comprises a relevance determination module configured to:
and determining the correlation degree of the demander and the information label according to the characteristic data of the demander and the characteristic data of the information label.
In one embodiment, the feature data includes feature vectors.
A relevance determination module configured to:
and calculating the inner product of the feature vector of the demand party and the feature vector of the information label to obtain the correlation degree of the demand party and the information label.
In one embodiment, the relevance determining module is configured to:
and determining the relevance of the target demander and each information tag so as to recommend the information tag configured to information delivery to the target demander, wherein the target demander is any demander.
In one embodiment, the feature data extraction module 940 is configured to:
and processing the vertex sequence by using the embedded model to obtain the characteristic data of at least one vertex.
In one embodiment, the association obtaining module 910 is configured to:
and screening out information release data of which release feedback data reach a preset standard, and acquiring the association relation between a plurality of demanders and a plurality of information labels from the information release data.
The specific details of each part in the above device have been described in detail in the method part embodiments, and thus are not described again.
Exemplary embodiments of the present disclosure also provide a computer-readable storage medium, which may be implemented in the form of a program product, including program code for causing an electronic device to perform the steps according to various exemplary embodiments of the present disclosure described in the above-mentioned "exemplary method" section of this specification, when the program product is run on the electronic device. In one embodiment, the program product may be embodied as a portable compact disc read only memory (CD-ROM) and include program code, and may be run on an electronic device, such as a personal computer. However, the program product of the present disclosure is not limited thereto, and in this document, a readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
A computer readable signal medium may include a propagated data signal with readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A readable signal medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Program code embodied on a readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Program code for carrying out operations for the present disclosure may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., through the internet using an internet service provider).
Exemplary embodiments of the present disclosure also provide an electronic device, which may be a background server of an information platform. The electronic device is explained below with reference to fig. 10. It should be understood that the electronic device 1000 shown in fig. 10 is only an example and should not bring any limitations to the function and scope of use of the embodiments of the present disclosure.
As shown in fig. 10, the electronic device 1000 is embodied in the form of a general purpose computing device. The components of the electronic device 1000 may include, but are not limited to: at least one processing unit 1010, at least one memory unit 1020, and a bus 1030 that couples various system components including the memory unit 1020 and the processing unit 1010.
Where the storage unit stores program code that may be executed by the processing unit 1010 to cause the processing unit 1010 to perform the steps according to various exemplary embodiments of the present invention described in the "exemplary methods" section above in this specification. For example, the processing unit 1010 may perform the method steps as shown in fig. 2, and the like.
The memory unit 1020 may include volatile memory units such as a random access memory unit (RAM)1021 and/or a cache memory unit 1022, and may further include a read only memory unit (ROM) 1023.
Storage unit 1020 may also include a program/utility 1024 having a set (at least one) of program modules 1025, such program modules 1025 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may comprise an implementation of a network environment.
Bus 1030 may include a data bus, an address bus, and a control bus.
The electronic device 1000 may also communicate with one or more external devices 1100 (e.g., keyboard, pointing device, bluetooth device, etc.), which may be through input/output (I/O) interfaces 1040. The electronic device 1000 may also communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network such as the internet) through the network adapter 1050. As shown, the network adapter 1050 communicates with the other modules of the electronic device 1000 via a bus 1030. It should be appreciated that although not shown, other hardware and/or software modules may be used in conjunction with the electronic device 1000, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, among others.
It should be noted that although in the above detailed description several modules or units of the device for action execution are mentioned, such a division is not mandatory. Indeed, the features and functions of two or more modules or units described above may be embodied in one module or unit, according to exemplary embodiments of the present disclosure. Conversely, the features and functions of one module or unit described above may be further divided into embodiments by a plurality of modules or units.
As will be appreciated by one skilled in the art, aspects of the present disclosure may be embodied as a system, method or program product. Accordingly, various aspects of the present disclosure may be embodied in the form of: an entirely hardware embodiment, an entirely software embodiment (including firmware, microcode, etc.) or an embodiment combining hardware and software aspects that may all generally be referred to herein as a "circuit," module "or" system. Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This application is intended to cover any variations, uses, or adaptations of the disclosure following, in general, the principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.
It will be understood that the present disclosure is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is to be limited only by the following claims.

Claims (14)

1. A method of feature extraction, comprising:
acquiring association relations between a plurality of demanders and a plurality of information labels from information delivery data;
establishing an information release relation graph based on the incidence relation, wherein the vertex of the information release relation graph comprises the demand party and the information label;
forming a plurality of vertex sequences through paths in the information delivery relation graph;
and extracting the characteristic data of at least one vertex according to the vertex sequence.
2. The method according to claim 1, wherein when obtaining the association relationship between a plurality of demanders and a plurality of information tags from information delivery data, the method further comprises:
acquiring the incidence relation between the information tag and the tag category;
the establishing of the information delivery relation graph based on the incidence relation comprises the following steps:
and establishing the information release relation graph based on the incidence relation between the demand party and the information label and the incidence relation between the information label and the label type, wherein the vertex of the information release relation graph comprises the demand party, the information label and the label type.
3. The method according to claim 2, wherein the path in the information delivery relationship graph is obtained by:
wandering in the information delivery relation graph according to a preset rule to acquire the path;
wherein the preset rule comprises at least one of:
if the current vertex is the demand side, the next vertex is an information label;
if the current vertex is an information label, the next vertex is a demand side;
if the previous vertex is a demand side and the current vertex is an information label, the next vertex is a label type;
if the previous vertex is of the label type and the current vertex is of the information label, the next vertex is a demand party;
the initial vertex of the path is a demand side;
the end vertex of the path is a demand side;
except for a starting vertex and an ending vertex, other vertexes in the path are all information labels or label categories;
the number of vertices in the path is within a predetermined range.
4. The method according to claim 1, wherein forming a plurality of vertex sequences through paths in the information placement relationship graph comprises:
and sequentially extracting homogeneous vertexes from the path to obtain the vertex sequence.
5. The method of claim 4, wherein said extracting homogenous vertices from said path in order to obtain said sequence of vertices comprises:
sequentially extracting vertexes corresponding to the demanders from the path to obtain a demander sequence;
and sequentially extracting vertexes corresponding to the information labels from the path to obtain an information label sequence.
6. The method of claim 5, wherein extracting feature data of at least one vertex from the sequence of vertices comprises:
obtaining characteristic data of at least one demander according to the demander sequence;
and obtaining the characteristic data of at least one information label according to the information label sequence.
7. The method of claim 6, further comprising:
and determining the correlation degree of the demander and the information label according to the characteristic data of the demander and the characteristic data of the information label.
8. The method of claim 7, wherein the feature data comprises a feature vector;
the determining the correlation degree of the demander and the information tag according to the characteristic data of the demander and the characteristic data of the information tag comprises the following steps:
and calculating the inner product of the feature vector of the demander and the feature vector of the information label to obtain the correlation degree of the demander and the information label.
9. The method of claim 7, wherein determining the correlation between the demander and the information tag according to the characteristic data of the demander and the characteristic data of the information tag comprises:
and determining the relevance between a target demand party and each information label so as to recommend the information label for information delivery to the target demand party, wherein the target demand party is any demand party.
10. The method of claim 1, wherein extracting feature data of at least one vertex from the sequence of vertices comprises:
and processing the vertex sequence by using an embedded model to obtain the characteristic data of the at least one vertex.
11. The method according to claim 1, wherein the obtaining the association relationship between the plurality of demanders and the plurality of information tags from the information delivery data comprises:
screening out information release data of which release feedback data reach a preset standard, and acquiring the association relation between a plurality of demanders and a plurality of information labels from the information release data.
12. A feature extraction device characterized by comprising:
the incidence relation acquisition module is configured to acquire incidence relations between a plurality of demanders and a plurality of information labels from the information delivery data;
the relation graph establishing module is configured to establish an information release relation graph based on the incidence relation, and the vertex of the information release relation graph comprises the demand party and the information label;
the vertex sequence extraction module is configured to form a plurality of vertex sequences through paths in the information delivery relation graph;
a feature data extraction module configured to extract feature data of at least one vertex from the sequence of vertices.
13. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the method of any one of claims 1 to 11.
14. An electronic device, comprising:
a processor; and
a memory for storing executable instructions of the processor;
wherein the processor is configured to perform the method of any of claims 1 to 11 via execution of the executable instructions.
CN202011457798.4A 2020-12-10 2020-12-10 Feature extraction method, feature extraction device, storage medium, and electronic apparatus Pending CN112488768A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011457798.4A CN112488768A (en) 2020-12-10 2020-12-10 Feature extraction method, feature extraction device, storage medium, and electronic apparatus

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011457798.4A CN112488768A (en) 2020-12-10 2020-12-10 Feature extraction method, feature extraction device, storage medium, and electronic apparatus

Publications (1)

Publication Number Publication Date
CN112488768A true CN112488768A (en) 2021-03-12

Family

ID=74916470

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011457798.4A Pending CN112488768A (en) 2020-12-10 2020-12-10 Feature extraction method, feature extraction device, storage medium, and electronic apparatus

Country Status (1)

Country Link
CN (1) CN112488768A (en)

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017092602A1 (en) * 2015-12-04 2017-06-08 阿里巴巴集团控股有限公司 Method for screening information delivery user and server
US20180203918A1 (en) * 2017-01-17 2018-07-19 International Business Machines Corporation Automatic feature extraction from a relational database
US20180307733A1 (en) * 2016-09-22 2018-10-25 Tencent Technology (Shenzhen) Company Limited User characteristic extraction method and apparatus, and storage medium
CN111061979A (en) * 2018-10-17 2020-04-24 北京字节跳动网络技术有限公司 User label pushing method and device, electronic equipment and medium
CN111090783A (en) * 2019-12-18 2020-05-01 北京百度网讯科技有限公司 Recommendation method, device and system, graph-embedded wandering method and electronic equipment
CN111382190A (en) * 2020-03-04 2020-07-07 腾讯科技（深圳）有限公司 Object recommendation method and device based on intelligence and storage medium
CN111400613A (en) * 2020-03-17 2020-07-10 苏宁金融科技(南京)有限公司 Article recommendation method, device, medium and computer equipment
CN111666346A (en) * 2019-03-06 2020-09-15 京东数字科技控股有限公司 Information merging method, transaction query method, device, computer and storage medium

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017092602A1 (en) * 2015-12-04 2017-06-08 阿里巴巴集团控股有限公司 Method for screening information delivery user and server
US20180307733A1 (en) * 2016-09-22 2018-10-25 Tencent Technology (Shenzhen) Company Limited User characteristic extraction method and apparatus, and storage medium
US20180203918A1 (en) * 2017-01-17 2018-07-19 International Business Machines Corporation Automatic feature extraction from a relational database
CN111061979A (en) * 2018-10-17 2020-04-24 北京字节跳动网络技术有限公司 User label pushing method and device, electronic equipment and medium
CN111666346A (en) * 2019-03-06 2020-09-15 京东数字科技控股有限公司 Information merging method, transaction query method, device, computer and storage medium
CN111090783A (en) * 2019-12-18 2020-05-01 北京百度网讯科技有限公司 Recommendation method, device and system, graph-embedded wandering method and electronic equipment
CN111382190A (en) * 2020-03-04 2020-07-07 腾讯科技（深圳）有限公司 Object recommendation method and device based on intelligence and storage medium
CN111400613A (en) * 2020-03-17 2020-07-10 苏宁金融科技(南京)有限公司 Article recommendation method, device, medium and computer equipment

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
DJORDJE GLIGORIJEVIC等: "Modeling Mobile User Actions for Purchase Recommendation using Deep Memory Networks", SIGIR’18, pages 1021 - 1024 *
李晓蕾;王英涛;许国伟;武柯;: "基于用户访问轨迹的信息精准投放***", 信息技术, no. 07, pages 58 - 61 *

Similar Documents

Publication Publication Date Title
CN111538912B (en) Content recommendation method, device, equipment and readable storage medium
US20180342004A1 (en) Cumulative success-based recommendations for repeat users
CN111444428A (en) Information recommendation method and device based on artificial intelligence, electronic equipment and storage medium
CN107346496B (en) Target user orientation method and device
US11200241B2 (en) Search query enhancement with context analysis
US20090198671A1 (en) System and method for generating subphrase queries
US20100250335A1 (en) System and method using text features for click prediction of sponsored search advertisements
CN111242748B (en) Method, apparatus, and storage medium for recommending items to a user
Paulson et al. Efficient large-scale internet media selection optimization for online display advertising
KR20160057475A (en) System and method for actively obtaining social data
CN111461841A (en) Article recommendation method, device, server and storage medium
US11734728B2 (en) Method and apparatus for providing web advertisements to users
CN111144952A (en) Advertisement recommendation method, device, server and storage medium based on user interests
CN104685899A (en) Dynamic media segment pricing
CN109417644A (en) The income optimization launched across screen advertisement
Hwang et al. An algorithm for movie classification and recommendation using genre correlation
CN111344696B (en) System and method for evaluating advertisements
KR102409719B1 (en) System and method for managing group purchase
CN112348592A (en) Advertisement recommendation method and device, electronic equipment and medium
CN111488526A (en) Recommendation method and device
CN114885185B (en) Live broadcast room recommendation method, content recommendation method, terminal and storage medium
CN113672758B (en) Song list generation method, device, medium and computing equipment
CN112200215B (en) Label feature extraction method and device, storage medium and electronic equipment
US20160217490A1 (en) Automatic Computation of Keyword Bids For Pay-Per-Click Advertising Campaigns and Methods and Systems Incorporating The Same
US20230316106A1 (en) Method and apparatus for training content recommendation model, device, and storage medium

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