CN117151814A - Personalized commodity recommendation and real-time dynamic adjustment method - Google Patents

Personalized commodity recommendation and real-time dynamic adjustment method Download PDF

Info

Publication number
CN117151814A
CN117151814A CN202311122662.1A CN202311122662A CN117151814A CN 117151814 A CN117151814 A CN 117151814A CN 202311122662 A CN202311122662 A CN 202311122662A CN 117151814 A CN117151814 A CN 117151814A
Authority
CN
China
Prior art keywords
commodity
time
real
browsing
historical
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
CN202311122662.1A
Other languages
Chinese (zh)
Inventor
李洋
谢丹
李晓琦
刘贞强
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.)
Xi'an Chuangye Tianxia Network Technology Co ltd
Original Assignee
Xi'an Chuangye Tianxia Network 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 Xi'an Chuangye Tianxia Network Technology Co ltd filed Critical Xi'an Chuangye Tianxia Network Technology Co ltd
Priority to CN202311122662.1A priority Critical patent/CN117151814A/en
Publication of CN117151814A publication Critical patent/CN117151814A/en
Pending legal-status Critical Current

Links

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
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0601—Electronic shopping [e-shopping]
    • G06Q30/0631—Item recommendations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/60—Protecting data
    • G06F21/602—Providing cryptographic facilities or services
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/60—Protecting data
    • G06F21/64—Protecting data integrity, e.g. using checksums, certificates or signatures
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Security & Cryptography (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Business, Economics & Management (AREA)
  • Finance (AREA)
  • Accounting & Taxation (AREA)
  • Software Systems (AREA)
  • Computer Hardware Design (AREA)
  • General Health & Medical Sciences (AREA)
  • Bioethics (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Business, Economics & Management (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Development Economics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Marketing (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a personalized commodity recommendation and real-time dynamic adjustment method, which relates to the technical field of electronic commerce, and comprises the steps of obtaining a commodity historical purchasing record, a commodity historical browsing record and a current real-time operation record of a user, and respectively generating a corresponding historical transaction data set, a historical interaction data set and a real-time data set; constructing a user portrait according to the historical transaction data set and the historical interaction data set, acquiring a user-associated custom keyword extended user portrait, and further generating a personalized recommendation model to screen out trend commodities; and positioning a plurality of real-time interactive commodities according to the real-time data set, and acquiring the content similarity among the plurality of real-time interactive commodities so as to reset the tendency commodities, thereby realizing personalized commodity recommendation and real-time dynamic adjustment of the commodities.

Description

Personalized commodity recommendation and real-time dynamic adjustment method
Technical Field
The invention relates to the technical field of electronic commerce, in particular to a personalized commodity recommendation and real-time dynamic adjustment method.
Background
With the rapid development of electronic commerce, more and more users face a great deal of trouble of commodity selection in the shopping process, and traditional commodity recommendation methods are mainly based on historical purchasing records and interest models of users, but have the problem that the methods cannot be adapted to the demands of the users in real time, so that the recommendation effect is poor.
How to analyze the behavior characteristics of the user according to the requirements of the user to construct corresponding user portraits, further achieve personalized commodity recommendation, monitor the requirements of the user in real time, and effectively conduct real-time dynamic adjustment of the recommended commodities in time so as to improve the accuracy of recommendation, further improve the shopping experience of the user, and the problems are all considered by the user.
Disclosure of Invention
In order to solve the above problems, the present invention is directed to providing a personalized commodity recommendation and real-time dynamic adjustment method.
The aim of the invention can be achieved by the following technical scheme: the personalized commodity recommendation and real-time dynamic adjustment method comprises the following steps:
step S1: acquiring commodity historical purchase records, commodity historical browsing records and current real-time operation records of a user, and respectively generating a corresponding historical transaction data set, a historical interaction data set and a real-time data set;
step S2: constructing a user portrait according to the historical transaction data set and the historical interaction data set, acquiring user-associated custom keywords, expanding the user portrait through the custom keywords, further generating a personalized recommendation model, and screening out trend commodities;
step S3: and positioning a plurality of real-time interactive commodities according to the real-time data set, acquiring the content similarity among the plurality of real-time interactive commodities, and further resetting the tendency commodities.
Further, the process of generating the historical transaction data set and the historical interaction data set includes:
setting a security database and a security monitoring period T, t=<T 1 ，T 1 `>，T 1 To monitor the start time, T 1 The' monitoring termination time;
the user performs operations on the commodity to generate different record information, wherein the record information comprises a commodity historical purchasing record, a commodity historical browsing record and a current real-time operation record, the record information is stored in a security database, the record information is backed up, and the record information is recorded in a T mode 1 Generating a security sequence at T 1 Receiving a safety sequence during the process, judging whether the safety sequence is tampered or not, and executing corresponding operation;
the commodity historical purchasing record comprises purchasing time, purchasing commodity information and purchasing times;
the commodity history browsing record comprises browsing behaviors and browsing time;
converting the commodity historical purchasing record and the commodity historical browsing record into corresponding binary 01 character strings, and summarizing a plurality of binary 01 character strings into a newly built text file, wherein the text file comprises a text file A and a text file B;
and storing the binary 01 character strings converted according to the commodity historical purchasing record into a text file A, storing the binary 01 character strings converted according to the commodity historical browsing record into a text file B, and summarizing a plurality of text files A and text files B to further respectively generate a historical transaction data set and a historical interaction data set.
Further, the process of generating the real-time data set according to the current real-time operation record includes:
acquiring a current real-time operation record, wherein the current real-time operation record comprises a plurality of operation items, the operation items comprise operation behaviors and operation commodities, a judging program is set, the operation behavior catalogue and the operation commodity list are input into the judging program in advance, and the operation items are input into the judging program for judgment:
when the operation behaviors corresponding to the operation items do not belong to the operation behavior catalogue, or the corresponding operation commodities are not in the operation commodity list, marking the corresponding operation items as error information, and eliminating the error information;
when the operation behaviors belong to the operation behavior catalogue, and the operation commodities are in the operation commodity list, marking the corresponding operation items as correct operation items, summarizing a plurality of correct operation items, converting the correct operation items into corresponding data streams, and further packaging the data streams into real-time data sets of corresponding users.
Further, the process of constructing the user portrait includes:
acquiring a historical transaction data set and a historical interaction data set, deconstructing a corresponding commodity historical purchasing record and commodity historical browsing record, and setting a high latitude characteristic drawing interval and a low latitude characteristic drawing interval;
acquiring purchase time, commodity information and number of purchases, and respectively marking as D 1 、D 2 And D 3 According to D 1 And D 3 Obtaining purchase frequency, marking as R, taking D 1 The corresponding value of two adjacent times of purchase time, and further the purchase interval, denoted as S, is obtained, and then r=s/D 3 ；
According to the purchase frequency R and the purchase goods information D 2 Forming a user characteristic sequence I, distributing corresponding characteristic weight coefficients I according to the numerical value of R to generate commodity drawing coefficients, acquiring browsing behaviors and browsing time, wherein the browsing behaviors comprise different behavior priorities, distributing corresponding characteristic weight coefficients II according to the behavior priorities to generate behavior drawing coefficients I, distributing corresponding characteristic weight coefficients III according to the browsing time to generate behavior drawing coefficients II;
and acquiring a first behavior drawing coefficient and a second behavior drawing coefficient corresponding to the same browsing behavior and browsing time, further acquiring the behavior drawing coefficients, establishing a Cartesian coordinate system according to the dependence relationship among the commodity drawing coefficients, the behavior drawing coefficients, the high latitude characteristic drawing interval and the low latitude characteristic drawing interval, acquiring a user coordinate set, and further constructing a user portrait.
Further, the process of expanding the user portrayal includes:
each user has a corresponding custom keyword, wherein the custom keyword comprises a plurality of commodity labels, and the commodity labels comprise a commodity price interval, commodity keywords and commodity modules;
the commodity price interval comprises an optimal trend price interval, an acceptable price interval and an unacceptable price interval;
the commodity keywords comprise commodity names, commodity good scores, commodity associated store information and commodity fuzzy associated words;
the commodity module is set as a priority display module, a secondary priority display module and a shielding module;
and taking the custom keywords as expansion items of the user portrait, and expanding the user portrait through the expansion items.
Further, the process of constructing the personalized recommendation model comprises the following steps:
acquiring a user portrait, and constructing a commodity recommendation model through a preset modeling language, wherein the personalized recommendation model is constructed according to different modeling parameters, and the modeling parameters comprise a first modeling parameter, a second modeling parameter and a third modeling parameter;
taking a commodity price interval as a first modeling parameter, setting different price weights for the first modeling parameter, taking a commodity keyword as a second modeling parameter, counting search frequency corresponding to commodity search text information corresponding to the second modeling parameter, taking a commodity module as a third modeling parameter, setting a corresponding first display coefficient, a second display coefficient and a third display coefficient for the third modeling parameter, and converting a commodity recommendation model into a personalized recommendation model according to the first modeling parameter, the second modeling parameter and the third modeling parameter.
Further, the process of screening the trended commodity comprises:
the personalized recommendation model comprises a plurality of model elements, namely a user image element, a commodity price interval element, a commodity keyword element and a commodity module element, wherein the different model elements are correspondingElement integration sections, wherein lambda is the element integration section corresponding to the commodity price section element, commodity keyword element and commodity module element 1 、λ 2 Lambda of 3 ；
Acquiring a plurality of commodities as input data, and generating output data of the commodities through a personalized recommendation model, wherein the output data comprises user portrait categories, commodity prices, commodity keywords and commodity display information;
if the user portrait category belongs to the user portrait element, sequentially determining the commodity price, commodity key word, commodity display information and lambda corresponding to the commodity 1 、λ 2 Lambda of 3 Generating commodity points of commodities, further generating commodity grades, and marking tendency commodities according to the commodity grades;
if the user portrait category does not belong to the user portrait element, no operation is performed.
Further, the process of resetting the trending commodity includes:
acquiring a real-time data set, further acquiring a plurality of real-time interactive commodities positioned by the real-time data set, further acquiring content similarity among the real-time interactive commodities, wherein the real-time interactive commodities are associated with corresponding interactive behaviors, and the interactive behaviors comprise clicking behaviors, browsing behaviors and ordering behaviors;
if the clicking time and the clicking frequency corresponding to the clicking behaviors of the two real-time interactive commodities are in a preset clicking behavior deviation value interval, the association degree Q1 is given, the browsing behaviors comprise browsing stay time and browsing page turning time, if the value corresponding to the browsing stay time exceeds a preset webpage stay threshold value, the browsing coefficient I is given, if the value corresponding to the browsing page turning time exceeds a preset page turning limit value, the browsing coefficient II is given, the association degree Q2 is given according to the browsing coefficient I and the browsing coefficient II, if the ordering operation is executed, the association degree Q3 is given, and otherwise, the association degree Q3 is not given; and summarizing the association degree Q1, the association degree Q2 and the association degree Q3, further generating commodity association degree, setting commodity reset coefficients, judging according to the commodity association degree and the commodity reset coefficients, and executing corresponding operation according to a judging result.
Compared with the prior art, the invention has the beneficial effects that:
1. in the stage of generating a historical transaction data set, a historical interaction data set and a real-time data set, a safety database and a safety monitoring period are set, a safety sequence is synchronously generated, whether the safety database is invaded or not is monitored by judging whether the safety sequence is tampered, when the safety database is invaded, data stored in the safety database are encrypted, and when the safety database is not invaded, the data are directly used, so that the safety of the data and the efficiency of data processing are improved to a certain extent;
2. the method comprises the steps of constructing a user portrait corresponding to a user through a historical transaction data set and a historical interaction data set, acquiring a user-associated custom keyword, expanding the user portrait through the custom keyword as an expansion item, enriching the dimension of the user portrait, wherein the optimal trend price interval and the acceptable price interval set in the commodity price interval are commodity prices acceptable to the user, serving as positive feedback dimensions of the user portrait, namely, the data base of follow-up commodity recommendation, and the unacceptable price interval serving as negative feedback dimensions of the user portrait, serving as an auxiliary means for commodity filtering, constructing a personalized recommendation model according to the user portrait, screening trend commodities, dynamically adjusting the trend commodities in real time through the real-time data set, enhancing recommendation accuracy to a certain extent, and improving shopping experience of the user.
Drawings
FIG. 1 is a flow chart of the present invention.
Detailed Description
As shown in fig. 1, the personalized commodity recommendation and real-time dynamic adjustment method comprises the following steps:
step S1: acquiring commodity historical purchase records, commodity historical browsing records and current real-time operation records of a user, and respectively generating a corresponding historical transaction data set, a historical interaction data set and a real-time data set;
step S2: constructing a user portrait according to the historical transaction data set and the historical interaction data set, acquiring user-associated custom keywords, expanding the user portrait through the custom keywords, further generating a personalized recommendation model, and screening out trend commodities;
step S3: and positioning a plurality of real-time interactive commodities according to the real-time data set, acquiring the content similarity among the plurality of real-time interactive commodities, and further resetting the tendency commodities.
It should be further noted that, in the implementation process, the process of generating the historical transaction data set and the historical interaction data set includes:
setting a security database and a security monitoring period, and recording the security monitoring period as T, wherein T=is given as the security monitoring period<T 1 ，T 1 `>Wherein T is 1 For the monitoring start time of the safety monitoring period, T 1 The monitoring termination time of the safety monitoring period;
the user performs the operation of the commodity on the commodity shopping platform, generates different record information through the operation, wherein the record information comprises a commodity historical purchasing record, a commodity historical browsing record and a current real-time operation record, stores the record information in a safe database, backs up the record information, and stores the record information in T 1 Generating a security sequence at T 1 Receiving the safety sequence and judging whether the safety sequence is tampered or not when the safety sequence is in T 1 And T 1 When the record information is inconsistent, the security sequence is tampered, the corresponding security database is invaded, at the moment, the backup record information is encrypted, and at the time of T 1 Transmitting to a preset area to be read in the security database when the security sequence is in T 1 And T 1 When the time is consistent, the security sequence is not tampered, and the original recorded information is directly transmitted to the area to be read;
setting a login unit and a verification unit, inputting shopping account numbers and corresponding passwords preset by different users into the login unit, generating a verification form through the login unit, acquiring the verification form and verifying the legality of the users by the verification unit, if the verification form does not carry virus information, legally allowing login to a commodity shopping platform, otherwise, prohibiting the corresponding users from logging in, marking the users as blacklist users, distributing a reading sequence for the legal users, and acquiring the record information of a region to be read through the reading sequence;
the commodity historical purchasing record comprises purchasing time, purchasing commodity information and purchasing times;
the commodity history browsing record comprises browsing behaviors and browsing time;
converting the commodity historical purchasing record and the commodity historical browsing record into corresponding binary 01 character strings, and summarizing a plurality of binary 01 character strings into a newly built text file, wherein the text file comprises a text file A and a text file B;
storing a binary 01 character string converted according to the commodity historical purchasing record into a text file A, respectively establishing a text index A-ID (identification) of purchasing time, purchasing commodity information and purchasing times, and correspondingly mapping the text index A-ID to a text area A included in the text file A 1 Text area A 2 And text area A 3 ；
Storing a binary 01 character string converted according to the commodity history browsing record into a text file B, respectively establishing a text index B-ID (identity) for browsing behaviors and browsing time, and correspondingly mapping the text index B-ID to a text region B included in the text file B 1 And text region B 2 ；
Converting the shopping account number and the corresponding password of each user into a binary sequence, taking the binary sequence as a user primary index of each user, summarizing text files A of a plurality of users to generate a historical transaction data set, and summarizing a plurality of corresponding text files B to generate a historical interaction data set;
it should be further noted that, in the implementation process, the process of generating the real-time data set according to the current real-time operation record includes:
acquiring a real-time operation record, wherein the real-time operation record comprises a plurality of operation items, and the operation items have corresponding operation behaviors and operation commodities;
setting a judging program, wherein legal operation behavior catalogues and operation commodity lists are recorded in advance in the judging program, and a plurality of operation items are input into the judging program for judgment:
when the operation behavior corresponding to the operation item does not belong to the operation behavior catalog or the corresponding operation commodity is not in the operation commodity list, marking the corresponding operation item as error information and eliminating the error information if at least one of the two conditions is met;
when the operation behaviors belong to the operation behavior catalogue, and the operation commodities are in the operation commodity list, marking the corresponding operation items as correct operation items, summarizing a plurality of correct operation items, converting the correct operation items into corresponding data streams, and further packaging the data streams into real-time data sets of corresponding users;
it should be further noted that, in the implementation process, the process of constructing the user portrait includes:
acquiring a historical transaction data set and a historical interaction data set, deconstructing a corresponding commodity historical purchasing record and commodity historical browsing record, setting a high latitude characteristic drawing interval and a low latitude characteristic drawing interval which are respectively marked as omega 1 And omega 2 ；
Acquiring the purchase time, the purchase commodity information and the purchase times in the commodity historical purchase record and respectively marking as D 1 、D 2 And D 3 According to D 1 And D 3 Acquiring purchase frequency, recording the purchase frequency as R, and taking D 1 The corresponding two adjacent values of the purchase time are respectively marked as D 11 And D 12 Further, a purchase interval, denoted S, is obtained, s=d 12 －D 11 Units of S: day, then there is r=s/D3;
according to the purchase frequency R and the purchase goods information D 2 Forming a first user characteristic sequence, and recording the first user characteristic sequence as C 1 C is then 1 ＝<R，D 2 >The characteristic weight coefficient I with the large value to the small value is distributed to the user from the large value to the small value according to the value of R, so that a commodity drawing coefficient is generated and marked as alpha;
acquiring browsing behaviors and browsing time in commodity historical browsing records and respectively marking the browsing behaviors and the browsing time as D 4 And D 5 The browsing behavior D 4 Comprising different behavior priorities, according to whichDifferent characteristic weight coefficients II are distributed according to the priority sequence of the levels, so that a behavior drawing coefficient I is generated, and the browsing time D is used for obtaining the behavior drawing coefficient 5 The time value of the (a) is distributed from small to large to the characteristic weight coefficient III with the corresponding value from small to large, so as to generate the behavior drawing coefficient II;
acquiring a behavior drawing coefficient I and a behavior drawing coefficient II corresponding to the same browsing behavior and browsing time, and further adding to generate a behavior drawing coefficient which is recorded as beta;
when alpha is E omega 1 When commodity drawing is performed, a commodity drawing high-dimensional coordinate is generated and is marked as P1, and when alpha is epsilon omega 2 When the commodity drawing low-dimensional coordinate is generated, the commodity drawing low-dimensional coordinate is marked as P2;
when beta is E omega 1 When the behavior is generated, the behavior is plotted into a high-dimensional coordinate, which is marked as P3, when beta is epsilon omega 2 When the method is used, generating a behavior drawing low-dimensional coordinate, and marking the coordinate as P4;
establishing a Cartesian coordinate system, sequentially mapping P1, P2, P3 and P4 to the Cartesian coordinate system, generating a user coordinate set, and constructing a user portrait according to the user coordinate set;
it should be further noted that, in the implementation process, the process of expanding the user portrait includes:
each user has a corresponding custom keyword, wherein the custom keyword comprises a plurality of commodity labels which are defined by the user, and each commodity label comprises a commodity price interval, a commodity keyword and a commodity module;
the commodity price interval comprises an optimal trend price interval, an acceptable price interval and an unacceptable price interval;
the commodity keyword is commodity searching text information customized by a user, and the commodity searching text information comprises commodity names, commodity good scores, commodity associated store information and commodity fuzzy associated words;
the commodity module is set as a priority display module, a secondary priority display module and a shielding module;
the user-defined keywords are used as expansion items of the user portrait, and the expansion items comprise expansion item I, expansion item II and expansion item III;
the corresponding relation between the expansion term and the custom keyword is as follows:
the commodity price interval corresponds to the expansion item I;
the commodity keyword corresponds to an expansion term II;
the commodity module corresponds to an expansion item III;
it should be noted that, the user portrait generated is expanded by using the user-defined keywords corresponding to the user as expansion terms, so that the dimension of the user portrait is enriched, wherein the "optimal trend price interval" and the "acceptable price interval" set in the "commodity price interval" are commodity prices acceptable to the user, and are used as positive feedback dimensions of the user portrait, namely data base of the follow-up commodity recommendation, and the "unacceptable price interval" is used as negative feedback dimensions of the user portrait, and can be used as an auxiliary means for commodity filtering;
it should be further noted that, in the implementation process, the process of screening the prone goods includes:
acquiring a user portrait, and constructing a commodity recommendation model by taking a preset UML language as a modeling language;
converting the commodity recommendation model into a personalized recommendation model according to modeling parameters corresponding to the expansion items;
the personalized recommendation model is constructed according to different modeling parameters, wherein the modeling parameters comprise a first modeling parameter, a second modeling parameter and a third modeling parameter;
setting different price weights for an optimal trend price interval, an acceptable price interval and an unacceptable price interval corresponding to the first modeling parameter by taking the commodity price interval as the first modeling parameter;
taking the commodity keywords as second modeling parameters, establishing a corresponding text set for commodity search text information corresponding to the second modeling parameters, counting search frequencies corresponding to the search text information, and associating the search frequencies with the text set;
taking the commodity module as a third modeling parameter, and respectively setting a first display coefficient, a second display coefficient and a third display coefficient for a first display module, a second display module and a shielding module corresponding to the third modeling parameter;
converting the commodity recommendation model into a personalized recommendation model according to the first modeling parameter, the second modeling parameter and the third modeling parameter;
it should be noted that, the values corresponding to the price weights of the "optimal trend price interval", "acceptable price interval" and "unacceptable price interval" decrease in order, and the magnitude relation of the first display coefficient, the second display coefficient and the third display coefficient decrease in order.
Acquiring a built personalized recommendation model, wherein the personalized recommendation model comprises four model elements, namely a user image element, a commodity price interval element, a commodity keyword element and a commodity module element;
the different model elements have corresponding element integration intervals, and the element integration intervals corresponding to the commodity price interval element, the commodity keyword element and the commodity module element are respectively marked as lambda 1 、λ 2 Lambda of 3 ；
Acquiring a plurality of commodities as input data, and generating output data of the commodities through a personalized recommendation model, wherein the output data comprises user portrait categories, commodity prices, commodity keywords and commodity display information;
if the user portrait category belongs to the user portrait element, sequentially determining the commodity price, commodity key word, commodity display information and lambda corresponding to the commodity 1 、λ 2 Lambda of 3 Generating commodity points of commodities, further generating commodity grades, and marking tendency commodities according to the commodity grades;
if the user portrait category does not belong to the user portrait element, no operation is performed;
the commodity price is Pr, the commodity keywords and commodity display information are data1 and data2, the commodity keywords data1 are provided with corresponding word frequency coefficients and are marked as Qt, and the commodity display information is provided with corresponding display priority value coefficients and is marked as Ya;
if Pr is epsilon lambda 1 Then Pr and lambda are sequentially combined 1 The corresponding optimal trend price interval, acceptable price interval and unacceptable price interval are compared, and commodity integration coefficients S1, S2 and S3 are respectively given;
if Qt epsilon lambda 2 Then Qt and lambda are sequentially added 2 Comparing the corresponding search frequencies, and respectively endowing the commodity class II integral coefficients H1, H2 and H3;
if Ya epsilon lambda 3 Acquiring values corresponding to the first display coefficient, the second display coefficient and the third display coefficient, respectively marking the values as C1, C2 and C3, and endowing the products with three types of integral coefficients of corresponding sizes according to the sequence of C1 > C2 > C3, namely G1, G2 and G3;
accumulating the commodity class-one integral coefficient, the commodity class-two integral coefficient and the commodity class-three integral coefficient to generate commodity integral, marking as a Shop-X, and acquiring commodity grade corresponding to the commodity according to the commodity integral;
setting a trend commodity recommendation interval, and marking the trend commodity recommendation interval as [ Min, max ];
if the Shop-X epsilon [ Min, max ], marking the commodity as a trend commodity, and obtaining a commodity grade of the trend commodity, wherein the commodity grade comprises a grade one, a grade two and a grade three, the grade one is greater than the grade two, the grade two is greater than the grade three, the commodity corresponding to the grade one and the grade two is transferred to a priority display module for display, and the commodity corresponding to the grade three is transferred to a secondary priority display module for display;
if it isMarking the commodity as a filtered commodity, and transferring the filtered commodity to a shielding module for storage;
it should be further noted that, in the implementation process, the process of generating the dynamic adjustment policy and resetting the tendency commodity includes:
acquiring a real-time data set, further acquiring a plurality of real-time interactive commodities positioned by the real-time data set, and further acquiring the content similarity between the real-time interactive commodities;
two real-time interactive commodities are taken for analysis every time, the real-time interactive commodities are associated with corresponding interactive behaviors, and the interactive behaviors comprise clicking behaviors, browsing behaviors and ordering behaviors;
if the clicking time and the clicking frequency corresponding to the clicking behaviors of the two real-time interactive commodities are within a preset clicking behavior deviation value interval, the relevance Q1 is given;
analyzing browsing behaviors of two real-time interactive commodities, wherein the browsing behaviors comprise browsing stay time and browsing page turning time, if a numerical value corresponding to the browsing stay time exceeds a preset webpage stay threshold, a browsing coefficient I is given, if the numerical value corresponding to the browsing page turning time exceeds a preset page turning limit value, a browsing coefficient II is given, the browsing coefficients I of the two real-time interactive commodities are recorded as V1 and V2 respectively, the browsing coefficients II are recorded as F1 and F2, and if V1/V2 epsilon V and F1/F2 epsilon F, a relevance Q2 is given, wherein V is the stay relevance of the two real-time interactive commodities, and F is the page turning relevance of the two real-time interactive commodities;
if the two real-time interactive commodities execute the ordering operation, the association degree Q3 is given, otherwise, the association degree Q3 is not given;
summarizing the association degree Q1, the association degree Q2 and the association degree Q3, further generating commodity association degree, marking as J, setting commodity resetting coefficient as K, comparing the commodity association degree with the commodity resetting coefficient, and further resetting the tendency commodity;
if J is more than or equal to K, resetting the tendency commodity, namely taking the current real-time interaction commodity as a new tendency commodity and pushing the new tendency commodity to a user position corresponding to the user image;
if J is less than K, resetting the tendency commodity is not carried out;
the operations corresponding to J is more than or equal to K and J is less than K are dynamic adjustment strategies;
the above embodiments are only for illustrating the technical method of the present invention and not for limiting the same, and it should be understood by those skilled in the art that the technical method of the present invention may be modified or substituted without departing from the spirit and scope of the technical method of the present invention.

Claims (8)

1. The personalized commodity recommendation and real-time dynamic adjustment method is characterized by comprising the following steps of:
step S1: acquiring commodity historical purchase records, commodity historical browsing records and current real-time operation records of a user, and respectively generating a corresponding historical transaction data set, a historical interaction data set and a real-time data set;
step S2: constructing a user portrait according to the historical transaction data set and the historical interaction data set, acquiring user-associated custom keywords, expanding the user portrait through the custom keywords, further generating a personalized recommendation model, and screening out trend commodities;
step S3: and positioning a plurality of real-time interactive commodities according to the real-time data set, acquiring the content similarity among the plurality of real-time interactive commodities, and further resetting the tendency commodities.
2. The personalized good recommendation and real-time dynamic adjustment method according to claim 1, wherein the process of generating the historical transaction data set and the historical interaction data set comprises:
setting a security database and a security monitoring period T, t=<T 1 ，T 1 `>，T 1 To monitor the start time, T 1 The' monitoring termination time;
the user performs operations on the commodity to generate different record information, wherein the record information comprises a commodity historical purchasing record, a commodity historical browsing record and a current real-time operation record, the record information is stored in a security database, the record information is backed up, and the record information is recorded in a T mode 1 Generating a security sequence at T 1 Receiving a safety sequence during the process, judging whether the safety sequence is tampered or not, and executing corresponding operation;
the commodity historical purchasing record comprises purchasing time, purchasing commodity information and purchasing times;
the commodity history browsing record comprises browsing behaviors and browsing time;
converting the commodity historical purchasing record and the commodity historical browsing record into corresponding binary 01 character strings, and summarizing a plurality of binary 01 character strings into a newly built text file, wherein the text file comprises a text file A and a text file B;
and storing the binary 01 character strings converted according to the commodity historical purchasing record into a text file A, storing the binary 01 character strings converted according to the commodity historical browsing record into a text file B, and summarizing a plurality of text files A and text files B to further respectively generate a historical transaction data set and a historical interaction data set.
3. The personalized good recommendation and real-time dynamic adjustment method according to claim 2, wherein the process of generating the real-time data set according to the current real-time operation record comprises:
acquiring a current real-time operation record, wherein the current real-time operation record comprises a plurality of operation items, the operation items comprise operation behaviors and operation commodities, a judging program is set, the operation behavior catalogue and the operation commodity list are input into the judging program in advance, and the operation items are input into the judging program for judgment:
when the operation behaviors corresponding to the operation items do not belong to the operation behavior catalogue, or the corresponding operation commodities are not in the operation commodity list, marking the corresponding operation items as error information, and eliminating the error information;
when the operation behaviors belong to the operation behavior catalogue, and the operation commodities are in the operation commodity list, marking the corresponding operation items as correct operation items, summarizing a plurality of correct operation items, converting the correct operation items into corresponding data streams, and further packaging the data streams into real-time data sets of corresponding users.
4. The personalized good recommendation and real-time dynamic adjustment method according to claim 3, wherein the process of constructing the user representation comprises:
acquiring a historical transaction data set and a historical interaction data set, deconstructing a corresponding commodity historical purchasing record and commodity historical browsing record, and setting a high latitude characteristic drawing interval and a low latitude characteristic drawing interval;
acquiring purchase time, commodity information and number of purchases, and respectively marking as D 1 、D 2 And D 3 According to D 1 And D 3 Obtaining purchase frequency, marking as R, taking D 1 The corresponding value of two adjacent times of purchase time, and further the purchase interval, denoted as S, is obtained, and then r=s/D 3 ；
According to the purchase frequency R and the purchase goods information D 2 Forming a user characteristic sequence I, distributing corresponding characteristic weight coefficients I according to the numerical value of R to generate commodity drawing coefficients, acquiring browsing behaviors and browsing time, wherein the browsing behaviors comprise different behavior priorities, distributing corresponding characteristic weight coefficients II according to the behavior priorities to generate behavior drawing coefficients I, distributing corresponding characteristic weight coefficients III according to the browsing time to generate behavior drawing coefficients II;
and acquiring a first behavior drawing coefficient and a second behavior drawing coefficient corresponding to the same browsing behavior and browsing time, further acquiring the behavior drawing coefficients, establishing a Cartesian coordinate system according to the dependence relationship among the commodity drawing coefficients, the behavior drawing coefficients, the high latitude characteristic drawing interval and the low latitude characteristic drawing interval, acquiring a user coordinate set, and further constructing a user portrait.
5. The personalized good recommendation and real-time dynamic adjustment method according to claim 4, wherein the process of expanding the user representation comprises:
each user has a corresponding custom keyword, wherein the custom keyword comprises a plurality of commodity labels, and the commodity labels comprise a commodity price interval, commodity keywords and commodity modules;
the commodity price interval comprises an optimal trend price interval, an acceptable price interval and an unacceptable price interval;
the commodity keywords comprise commodity names, commodity good scores, commodity associated store information and commodity fuzzy associated words;
the commodity module is set as a priority display module, a secondary priority display module and a shielding module;
and taking the custom keywords as expansion items of the user portrait, and expanding the user portrait through the expansion items.
6. The personalized good recommendation and real-time dynamic adjustment method according to claim 5, wherein the process of constructing the personalized recommendation model comprises:
acquiring a user portrait, and constructing a commodity recommendation model through a preset modeling language, wherein the personalized recommendation model is constructed according to different modeling parameters, and the modeling parameters comprise a first modeling parameter, a second modeling parameter and a third modeling parameter;
taking a commodity price interval as a first modeling parameter, setting different price weights for the first modeling parameter, taking a commodity keyword as a second modeling parameter, counting search frequency corresponding to commodity search text information corresponding to the second modeling parameter, taking a commodity module as a third modeling parameter, setting a corresponding first display coefficient, a second display coefficient and a third display coefficient for the third modeling parameter, and converting a commodity recommendation model into a personalized recommendation model according to the first modeling parameter, the second modeling parameter and the third modeling parameter.
7. The personalized good recommendation and real-time dynamic adjustment method according to claim 6, wherein the process of screening the trended good comprises:
the personalized recommendation model comprises a plurality of model elements, namely a user image element, a commodity price interval element, a commodity keyword element and a commodity module element, wherein different model elements have corresponding element integration intervals, and commodity price interval element and commodity keyword element are respectively recordedThe element integration interval corresponding to the commodity module element is lambda 1 、λ 2 Lambda of 3 ；
Acquiring a plurality of commodities as input data, and generating output data of the commodities through a personalized recommendation model, wherein the output data comprises user portrait categories, commodity prices, commodity keywords and commodity display information;
if the user portrait category belongs to the user portrait element, sequentially determining the commodity price, commodity key word, commodity display information and lambda corresponding to the commodity 1 、λ 2 Lambda of 3 Generating commodity points of commodities, further generating commodity grades, and marking tendency commodities according to the commodity grades;
if the user portrait category does not belong to the user portrait element, no operation is performed.
8. The personalized good recommendation and real-time dynamic adjustment method according to claim 7, wherein the process of resetting the trending good comprises:
acquiring a real-time data set, further acquiring a plurality of real-time interactive commodities positioned by the real-time data set, further acquiring content similarity among the real-time interactive commodities, wherein the real-time interactive commodities are associated with corresponding interactive behaviors, and the interactive behaviors comprise clicking behaviors, browsing behaviors and ordering behaviors;
if the clicking time and the clicking frequency corresponding to the clicking behaviors of the two real-time interactive commodities are in a preset clicking behavior deviation value interval, the association degree Q1 is given, the browsing behaviors comprise browsing stay time and browsing page turning time, if the value corresponding to the browsing stay time exceeds a preset webpage stay threshold value, the browsing coefficient I is given, if the value corresponding to the browsing page turning time exceeds a preset page turning limit value, the browsing coefficient II is given, the association degree Q2 is given according to the browsing coefficient I and the browsing coefficient II, if the ordering operation is executed, the association degree Q3 is given, and otherwise, the association degree Q3 is not given; and summarizing the association degree Q1, the association degree Q2 and the association degree Q3, further generating commodity association degree, setting commodity reset coefficients, judging according to the commodity association degree and the commodity reset coefficients, and executing corresponding operation according to a judging result.
CN202311122662.1A 2023-09-01 2023-09-01 Personalized commodity recommendation and real-time dynamic adjustment method Pending CN117151814A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311122662.1A CN117151814A (en) 2023-09-01 2023-09-01 Personalized commodity recommendation and real-time dynamic adjustment method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311122662.1A CN117151814A (en) 2023-09-01 2023-09-01 Personalized commodity recommendation and real-time dynamic adjustment method

Publications (1)

Publication Number Publication Date
CN117151814A true CN117151814A (en) 2023-12-01

Family

ID=88903991

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311122662.1A Pending CN117151814A (en) 2023-09-01 2023-09-01 Personalized commodity recommendation and real-time dynamic adjustment method

Country Status (1)

Country Link
CN (1) CN117151814A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117474636A (en) * 2023-12-27 2024-01-30 广州宇中网络科技有限公司 Platform user recommendation method and system based on big data
CN117575747A (en) * 2024-01-19 2024-02-20 山东街景智能制造科技股份有限公司 Personalized recommendation method based on user analysis
CN117688250A (en) * 2024-02-04 2024-03-12 国网湖北省电力有限公司信息通信公司 Unified data dynamic service management system and method suitable for electric power full scene

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117474636A (en) * 2023-12-27 2024-01-30 广州宇中网络科技有限公司 Platform user recommendation method and system based on big data
CN117474636B (en) * 2023-12-27 2024-04-12 广州宇中网络科技有限公司 Platform user recommendation method and system based on big data
CN117575747A (en) * 2024-01-19 2024-02-20 山东街景智能制造科技股份有限公司 Personalized recommendation method based on user analysis
CN117575747B (en) * 2024-01-19 2024-04-05 山东街景智能制造科技股份有限公司 Personalized recommendation method based on user analysis
CN117688250A (en) * 2024-02-04 2024-03-12 国网湖北省电力有限公司信息通信公司 Unified data dynamic service management system and method suitable for electric power full scene
CN117688250B (en) * 2024-02-04 2024-04-16 国网湖北省电力有限公司信息通信公司 Unified data dynamic service management system and method suitable for electric power full scene

Similar Documents

Publication Publication Date Title
CN117151814A (en) Personalized commodity recommendation and real-time dynamic adjustment method
US6647390B2 (en) System and methods for standardizing data for design review comparisons
US8180713B1 (en) System and method for searching and identifying potential financial risks disclosed within a document
CN109767318A (en) Loan product recommended method, device, equipment and storage medium
CN103377190B (en) Trading platform based supplier information searching method and device
CN111444944A (en) Information screening method, device, equipment and storage medium based on decision tree
US20090216696A1 (en) Determining relevant information for domains of interest
CN113706251B (en) Model-based commodity recommendation method, device, computer equipment and storage medium
WO2009137788A2 (en) Legal instrument management platform with transaction management
US11921737B2 (en) ETL workflow recommendation device, ETL workflow recommendation method and ETL workflow recommendation system
US20140007261A1 (en) Business application search
CN116757808A (en) Automatic bidding document generation method and system based on big data
US20230359599A1 (en) Method and system for managing metadata
Chalyi et al. Method of constructing explanations for recommender systems based on the temporal dynamics of user preferences
WO2005065392A2 (en) System and method for adaptive decision making analysis and assessment
CN114861050A (en) Feature fusion recommendation method and system based on neural network
CN115374354A (en) Scientific and technological service recommendation method, device, equipment and medium based on machine learning
Yudowati et al. Big data framework for auditing process
EP1814048A2 (en) Content analytics of unstructured documents
US8887045B2 (en) System and method for providing data links
CN115982429B (en) Knowledge management method and system based on flow control
Edi Surya et al. Recommendation System with Content-Based Filtering in NFT Marketplace
Chiang et al. The cyclic model analysis on sequential patterns
Cui et al. An online book recommendation system based on web service
CN115907968A (en) Wind control rejection inference method and device based on pedestrian credit

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication