CN113487379A - Product recommendation method and device based on conversation and electronic equipment - Google Patents
Product recommendation method and device based on conversation and electronic equipment Download PDFInfo
- Publication number
- CN113487379A CN113487379A CN202110707740.9A CN202110707740A CN113487379A CN 113487379 A CN113487379 A CN 113487379A CN 202110707740 A CN202110707740 A CN 202110707740A CN 113487379 A CN113487379 A CN 113487379A
- Authority
- CN
- China
- Prior art keywords
- vector
- target user
- product
- attribute
- feedback
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 41
- 239000013598 vector Substances 0.000 claims abstract description 416
- 230000004044 response Effects 0.000 claims abstract description 15
- 239000000047 product Substances 0.000 claims description 200
- 230000003993 interaction Effects 0.000 claims description 16
- 230000006870 function Effects 0.000 claims description 12
- 239000000706 filtrate Substances 0.000 claims description 6
- 238000012790 confirmation Methods 0.000 claims description 4
- 230000000694 effects Effects 0.000 abstract description 10
- 238000010586 diagram Methods 0.000 description 10
- 238000012545 processing Methods 0.000 description 9
- 230000006399 behavior Effects 0.000 description 7
- 238000005070 sampling Methods 0.000 description 5
- 238000004590 computer program Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 230000014759 maintenance of location Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 235000011888 snacks Nutrition 0.000 description 1
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
- G06Q30/06—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
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Mathematical Physics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Artificial Intelligence (AREA)
- Human Computer Interaction (AREA)
- Development Economics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to the technical field of computers, in particular to a product recommendation method and device based on a conversation mode and electronic equipment, wherein the product recommendation method and device based on the conversation mode comprise the following steps: acquiring a target user vector; performing vector matching on the target user vector, and determining a product vector or an attribute vector matched with the target user vector in a candidate pool; determining an inquiry statement in a dialogue with a target user, and accepting feedback in response to the inquiry statement; and updating the candidate pool, the target user interest pool and the target user vector according to the feedback. The invention can quickly and accurately position the product requirements of the user, carry out accurate recommendation and increase the purchase satisfaction of the digital products of new users or low-activity users.
Description
Technical Field
The invention relates to the technical field of computers, in particular to a product recommendation method and device based on a conversation mode and electronic equipment.
Background
When facing hundreds of digital products, users often do not know which is suitable for themselves, and the industry often recommends products to users through recommendation technologies, wherein the recommendation technologies all depend on historical interaction data of the users. The method is characterized in that interactive information of a new user or a low-activity user is almost not available, a traditional recommendation technology cannot be used, and in order to solve the problem that the new user recommends cold start, digital products with a large number of purchasers are recommended through a statistical means based on historical records of other users in the industry, or products are recommended to the user randomly in order to increase diversity. Both methods are difficult to accurately locate the digital product requirements of new users, so that the users directly quit purchasing pages, or after purchasing digital products recommended by the system, the users find that the digital products are not needed by themselves and return to insurance, which directly or indirectly leads to user loss.
Disclosure of Invention
The invention provides a product recommendation method and device based on a conversation mode and electronic equipment, which can quickly and accurately position the product requirements of users, carry out accurate recommendation, increase the purchase satisfaction of digital products of new users or low-activity users, increase the number of daily active users and improve the retention rate.
An embodiment of the present specification provides a product recommendation method based on a dialog formula, including:
acquiring a target user vector;
carrying out vector matching on the target user vector through the attribute vector in the target user interest pool, and determining a product vector or an attribute vector matched with the target user vector in a candidate pool;
determining an inquiry statement when the target user carries out conversation according to the product vector or the attribute vector matched with the target user vector, and receiving feedback responding to the inquiry statement;
when the feedback is positive feedback and the response is attribute vector, adding the attribute vector to a target user interest pool, filtering product vectors which do not comprise the attribute vector in a candidate pool, updating the target user vector according to the feedback, and executing vector matching and inquiring again;
when the feedback is positive feedback and the response is a product vector, recommending a product corresponding to the product vector to the target user, and ending;
when the feedback is negative feedback, removing the corresponding attribute vector or product vector in the candidate pool, updating the target user vector according to the feedback, and performing vector matching and querying again.
Preferably, the obtaining the target user vector includes:
acquiring historical user interaction data;
based on the characteristic relevance of the historical user interaction data, a historical user vector, a product vector and an attribute vector are respectively constructed, wherein the historical user vector is a target user vector for carrying out conversation with a target user for the first time.
Preferably, the vector matching of the target user vector includes:
and pairing the target user vector and the product vector and/or the attribute vector in the candidate pool in sequence through the attribute vector in the target user interest pool to obtain a plurality of pairing combinations.
Preferably, the determining a product vector or an attribute vector in the candidate pool that matches the target user vector includes:
and respectively calculating product return values of a plurality of pairing combinations based on a preset return function, wherein a product vector or an attribute vector corresponding to the pairing combination corresponding to the product return value with the largest numerical value is a product vector or an attribute vector matched with the target user vector.
Preferably, the determining the query statement in the dialog with the target user includes:
when the attribute vector is matched with the target user vector, carrying out preference inquiry of corresponding attributes on the target user in the next round of conversation;
and when the product vector is matched with the target user vector, recommending the product to the target user in the next conversation.
Preferably, the updating the target user vector according to the feedback includes:
initializing a target user vector obeying multi-dimensional Gaussian distribution, wherein the target user vector comprises a new user vector and/or an inactive user vector;
acquiring the current number of conversation turns of the target user;
when the current number of turns is lower than a preset number of turns threshold, updating parameters which are distributed from a multidimensional Gaussian in a target user vector by responding to feedback of the query statement to obtain an updated target user vector;
and when the current conversation turn number is higher than a preset conversation turn number threshold value, ending the conversation.
An embodiment of the present specification further provides a product recommendation device based on a dialog style, including:
the vector acquisition module is used for acquiring a target user vector;
the vector matching module is used for carrying out vector matching on the target user vector through the attribute vector in the target user interest pool and determining a product vector or an attribute vector matched with the target user vector in the candidate pool;
the statement confirmation module is used for determining an inquiry statement when the target user carries out conversation according to the product vector or the attribute vector matched with the target user vector and receiving feedback responding to the inquiry statement;
the first execution module is used for adding the attribute vector to a target user interest pool when the feedback is positive feedback and the response is the attribute vector, filtering product vectors which do not comprise the attribute vector in a candidate pool, updating the target user vector according to the feedback, and executing vector matching and inquiring again;
the second execution module recommends a product corresponding to the product vector to the target user when the feedback is positive feedback and the response is the product vector, and the operation is finished;
and the third execution module is used for removing the corresponding attribute vector or product vector in the candidate pool when the feedback is negative feedback, updating the target user vector according to the feedback, and executing vector matching and inquiry again.
Preferably, the determining a product vector or an attribute vector in the candidate pool that matches the target user vector includes:
and respectively calculating product return values of a plurality of pairing combinations based on a preset return function, wherein a product vector or an attribute vector corresponding to the pairing combination corresponding to the product return value with the largest numerical value is a product vector or an attribute vector matched with the target user vector.
An electronic device, wherein the electronic device comprises:
a processor and a memory storing computer executable instructions that, when executed, cause the processor to perform the method of any of the above.
A computer readable storage medium, wherein the computer readable storage medium stores one or more programs which, when executed by a processor, implement the method of any of the above.
The beneficial effects are that:
the invention adopts a conversational recommendation method, quickly and accurately positions the product requirements of the user by updating the target user vector, carries out accurate recommendation, increases the purchase satisfaction of the digital products of new users or low-activity users, brings the increase of the number of daily active users, and improves the retention rate; meanwhile, a Thompson sampling mechanism is adopted, the user interest is updated after each round of conversation on the premise that the target user is disturbed as little as possible, the target user vector is updated according to the conversation feedback, the corresponding communication statement is adjusted, and the optimal product recommendation opportunity is finally determined.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the application and together with the description serve to explain the application and not to limit the application. In the drawings:
FIG. 1 is a schematic diagram illustrating a dialog-based product recommendation method according to an embodiment of the present disclosure;
FIG. 2 is a schematic structural diagram of a dialog-based product recommendation device according to an embodiment of the present disclosure;
fig. 3 is a schematic structural diagram of an electronic device provided in an embodiment of the present disclosure;
fig. 4 is a schematic diagram of a computer-readable medium provided in an embodiment of the present specification.
Detailed Description
Exemplary embodiments of the present invention 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 embodiments set forth herein. Rather, these exemplary embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the invention to those skilled in the art. The same reference numerals denote the same or similar elements, components, or parts in the drawings, and thus their repetitive description will be omitted.
Features, structures, characteristics or other details described in a particular embodiment do not preclude the fact that the features, structures, characteristics or other details may be combined in a suitable manner in one or more other embodiments in accordance with the technical idea of the invention.
In describing particular embodiments, the present invention has been described with reference to features, structures, characteristics or other details that are within the purview of one skilled in the art to provide a thorough understanding of the embodiments. One skilled in the relevant art will recognize, however, that the invention may be practiced without one or more of the specific features, structures, characteristics, or other details.
The diagrams depicted in the figures are exemplary only, and do not necessarily include all of the contents and operations/steps, nor do they necessarily have to be performed in the order depicted. For example, some operations/steps may be decomposed, and some operations/steps may be combined or partially combined, so that the actual execution sequence may be changed according to the actual situation.
The block diagrams shown in the figures are functional entities only and do not necessarily correspond to physically separate entities. I.e. 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 means and/or microcontroller means.
The term "and/or" and/or "includes all combinations of any one or more of the associated listed items.
Referring to fig. 1, a schematic diagram of a dialog-based product recommendation method provided in an embodiment of the present disclosure includes:
s101: acquiring a target user vector;
in a preferred embodiment of the invention, the target user vector acquired for the first time is from active historical user interaction data, an FM algorithm is adopted to train the model, then the characteristics of the active historical user interaction data are extracted, the characteristic data are input into the model, and finally the historical user vector, the product vector and the attribute vector are output, wherein the historical user vector at the moment is used as the target user vector used before the conversation with the target client for the first time; the product vector is a vector formed by feature data of a product used by a historical user, for example, a vector formed by data such as a name of the product, a price of the product, a type of the product, and a duration of the product, and specifically, for example, the product vector of a digital product may be a name of the digital product, a product age, a type of the digital product, a price of the digital product, and the like; the attribute vector is a vector formed by attribute information of products related to the product vector, for example, data in a part of the product vector forms the attribute vector to communicate with a user, specifically, the attribute vector may be a type of a digital product, the user is inquired about a type of a required digital product through the attribute vector, so as to filter a type of the digital product desired by the user, and based on the type, different attribute vectors are continuously matched to complete the filtering of the product, and finally inquiry information corresponding to the filtered product is sent to the user, so that the product is recommended to the user through interaction. When a next round of conversation needs to be carried out on the target client, updating the target user vector according to information fed back by the conversation content of the previous round; when the target user vector is updated, the characteristic parameters or the group of the user finally obey multi-dimensional Gaussian distribution
So the historical user vector is initialized using equation (1):
wherein, muuIs a target user vector, uinitFor initializing the obtained historical user vector, the historical user vector is expressed as U ═ U1，u2，...，uNN is the number of active historical users, BuInitialized to the identity matrix and set an auxiliary parameter fu＝uinit。
Then B in the algorithm is paired through feedback, product vectors and attribute vectors of the target useruAnd fuAdjusting to obtain new multidimensional Gaussian distribution
Sampling is carried out based on new multi-dimensional Gaussian distribution to obtain a subsequent target user vector, the digital product demand of a user is quickly and accurately positioned based on the subsequent target user vector, accurate recommendation is carried out, the purchase satisfaction degree of the digital product of a new user or a low-activity user is increased, and the problem of cold start of the new user and an inactive user is solved.
In this embodiment, the digital product may be a commodity of online shopping, such as clothes, shoes, snacks, furniture, tools, etc., an insurance product, digital currency, or a fund product, which is not described herein again.
S102: carrying out vector matching on the target user vector through the attribute vector in the target user interest pool, and determining a product vector or an attribute vector matched with the target user vector in a candidate pool;
in a preferred embodiment of the present invention, the target user vector and the product vectors and/or the attribute vectors in the candidate pool are sequentially paired through the attribute vector in the target user interest pool to obtain a plurality of pairing combinations, then the vector corresponding to each pairing combination is input into a predefined reward function to calculate the product reward value corresponding to each pairing combination, the pairing combination corresponding to the maximum product reward value is the product vector or the attribute vector matching the target user vector in the determined candidate pool, and the maximum product reward value is calculated by the above method to determine the best recommendation time. The maximum product return value a (t) is calculated by using formula (2), as follows:
wherein, a (t) is the maximum return value, argmax represents the maximum value of the function value range, A is the candidate pool set, xaIs a product vector or an attribute vector, P, in a candidate pooluIs a user interest pool, piIs a vector of attributes of interest to a user in the user interest pool,
is muuThe transposing of (1).
S103: determining an inquiry statement when the target user carries out conversation according to the product vector or the attribute vector matched with the target user vector, and receiving feedback responding to the inquiry statement;
in the preferred embodiment of the present invention, if the pairing with the largest product return value comes from the pairing of the target user vector and a certain attribute vector, the behavior of the dialog system in this round is to ask the user's preference for the attribute; if the pairing with the maximum product return value comes from the pairing of the target user vector and a certain product vector, the product is recommended to the target user by the dialog system behavior in the current round, and meanwhile, the feedback given by the target user is responded to the system behavior. By the method, the digital product requirements of new users or low-activity users can be quickly and accurately positioned, and accurate recommendation can be carried out.
S104: when the feedback is positive feedback and the response is attribute vector, adding the attribute vector to a target user interest pool, filtering product vectors which do not comprise the attribute vector in a candidate pool, updating the target user vector according to the feedback, and executing vector matching and inquiring again;
in a preferred embodiment of the present invention, if the current dialog wheel queries the target user for the attribute, and the target user gives positive feedback on the current attribute, the attribute vector corresponding to the attribute is added to the target user interest pool, meanwhile, the product vector not containing the attribute vector is filtered from the candidate pool, then the target user vector is updated according to the user feedback, the product vector, the attribute vector, and the like, and S102-S103 is continuously executed until the product recommendation is successful or the number of dialog wheels exceeds the preset threshold, and the dialog is stopped. The product vector and the attribute vector quality inspection are associated, and products more suitable for target users can be further recommended through confirmation of the attribute vector. By the method, the digital product requirements of the users can be quickly and accurately positioned, accurate recommendation is carried out, and the purchase satisfaction of the digital products of new users or low-activity users is increased.
S105: when the feedback is positive feedback and the response is a product vector, recommending a product corresponding to the product vector to the target user, and ending;
in the preferred embodiment of the invention, if the product is recommended to the target user and the target user accepts the recommended product, the conversation process for the target client is ended, and the current task is completed, so that the method reduces excessive disturbance to the target user.
S106: when the feedback is negative feedback, removing the corresponding attribute vector or product vector in the candidate pool, updating the target user vector according to the feedback, and performing vector matching and querying again.
In a preferred embodiment of the present invention, if the target user gives negative feedback to the recommended product vector or the queried attribute vector, the corresponding product vector and attribute vector are removed from the candidate pool, and then the target user vector is updated according to the user feedback, product vector, attribute vector, etc., and S102-S103 are continuously executed until the product recommendation is successful or the number of dialog turns exceeds the preset threshold, and the dialog is stopped. By the method, the digital product requirements of the users can be quickly and accurately positioned, accurate recommendation is carried out, and the purchase satisfaction of the digital products of new users or low-activity users is increased.
Further, the obtaining a target user vector includes:
acquiring historical user interaction data;
based on the characteristic relevance of the historical user interaction data, a historical user vector, a product vector and an attribute vector are respectively constructed, wherein the historical user vector is a target user vector for carrying out conversation with a target user for the first time.
In the preferred embodiment of the invention, historical user interaction data is obtained, then the vector model is trained by adopting an FM algorithm, then the active historical user interaction data is subjected to feature extraction and is input into the vector model, and finally, a historical user vector, a product vector and an attribute vector are output, wherein the historical user vector at the moment is used as a target user vector used before the first conversation with a target client.
Further, the vector matching the target user vector includes:
and pairing the target user vector and the product vector and/or the attribute vector in the candidate pool in sequence through the attribute vector in the target user interest pool to obtain a plurality of pairing combinations.
In a preferred embodiment of the invention, the target user vector and the product vector in the candidate pool are sequentially paired through the attribute vector in the target user interest pool, and the target user vector and the attribute vector in the candidate pool are sequentially paired to obtain a plurality of pairing combinations, and the inquiry mode for the user is determined in the form of the pairing combinations, so that the digital product requirement of the user is quickly and accurately positioned.
Further, the determining a product vector or an attribute vector in the candidate pool that matches the target user vector includes:
and respectively calculating product return values of a plurality of pairing combinations based on a preset return function, wherein a product vector or an attribute vector corresponding to the pairing combination corresponding to the product return value with the largest numerical value is a product vector or an attribute vector matched with the target user vector.
In a preferred embodiment of the present invention, product return values of a plurality of pairing combinations are respectively calculated through a preset return function, the calculated product return values are counted to find a maximum product return value, a product vector or an attribute vector corresponding to the pairing combination corresponding to the maximum product return value is a product vector or an attribute vector matched with the target user vector, a corresponding query statement is provided to a user based on the matching combination, and an optimal recommendation time of a product is determined by calculating the maximum product return value, so that a digital product requirement of a new user or a low-activity user is quickly and accurately positioned, and accurate recommendation is performed.
Further, the determining the query statement in the dialog with the target user includes:
when the attribute vector is matched with the target user vector, carrying out preference inquiry of corresponding attributes on the target user in the next round of conversation;
and when the product vector is matched with the target user vector, recommending the product to the target user in the next conversation.
In the preferred embodiment of the present invention, if the pairing with the largest product return value comes from the pairing of the target user vector and a certain attribute vector, the behavior of the dialog system in this round is to ask the user's preference for the attribute vector; if the pairing with the maximum product return value comes from the pairing of a target user vector and a certain product vector, the behavior of the dialog system in the current round is to recommend the product corresponding to the product vector to the target user, and the method can quickly and accurately position the digital product requirements of the new user or the low-activity user to carry out accurate recommendation.
Further, the updating the target user vector according to the feedback includes:
initializing a target user vector obeying multi-dimensional Gaussian distribution, wherein the target user vector comprises a new user vector and/or an inactive user vector;
acquiring the current number of conversation turns of the target user;
when the current number of turns is lower than a preset number of turns threshold, updating parameters which are distributed from a multidimensional Gaussian in a target user vector by responding to feedback of the query statement to obtain an updated target user vector;
and when the current conversation turn number is higher than a preset conversation turn number threshold value, ending the conversation.
In the preferred embodiment of the present invention, after the first wheel session is over, the system will determine B in the algorithm based on the feedback of the target user, the product vector and the attribute vector pairuAnd fuMake adjustments, wherein B is updateduAnd fuThe formula (3), (4), (5) and (6):
fu＝fu+r′a(t)*xa(t) (5)
wherein r isa(t)Giving feedback, r ', to the target user'a(t)Is according to ra(t)Calculated feedback adjustment parameter, PuFor the user interest pool, xa(t)For the product vectors or attribute vectors in the candidate pool after the current t-round of conversation,
is xa(t)Transpose of (P)uIs a target user interest pool, piIs the attribute vector of interest to the user in the target user interest pool.
Then according to pair BuAnd fuPerforming an update to obtain an updated multidimensional Gaussian distribution
Finally, sampling the target user vector according to the updated multidimensional Gaussian distribution to obtain an updated target user vector muu. Using the Thompson sampling mechanism, the target user vector is updated for each session, so that the control systemThe number of times of questioning is unified, and the optimal recommendation opportunity is determined by calculating the maximum return on the premise of disturbing the user as little as possible; meanwhile, by adopting the conversational recommendation method, the requirements of new users on digital products can be quickly and accurately positioned, and accurate recommendation is carried out.
In the preferred embodiment of the present invention, step 1, an offline initialization is performed, the product vector and the attribute vector in the initialization process are all vectors in the candidate pool, the historical user vector in the initialization process is data used for realizing the first content transmission with the user, and after the subsequent user has feedback, mu is constructed according to the feedback and information of the useruTo determine a subsequent dialog; step 2, sampling is carried out based on the updated multidimensional Gaussian distribution to obtain a target user vector under the current conversation wheel; step 3, selecting corresponding strategy behaviors based on the pairing combination corresponding to the calculated maximum product return value; step 4, if the number T of the conversation turns exceeds a set conversation turn threshold value T, directly ending the conversation; otherwise, performing step 5; step 5, after the system behavior is executed, the user gives feedback ra(t)Updating the candidate pool A and the user interest pool P based on the user feedbackuAnd the user embedding vector muu。
Step 1 may refer to the above embodiment for "obtaining historical user interaction data; based on the relevance of features in the historical user interaction data, a historical user vector, a product vector and an attribute vector are respectively constructed, wherein the historical user vector is a target user vector for carrying out conversation with a target user for the first time. "step 2 may refer to the above embodiment to" initialize target user vectors subject to multidimensional gaussian distribution, where the target user vectors include new user vectors, and/or inactive user vectors; acquiring the current number of conversation turns of the target user; when the current number of turns is lower than a preset number of turns threshold, updating parameters which are distributed from a multidimensional Gaussian in a target user vector by responding to feedback of the query statement to obtain an updated target user vector; "with reference to the above embodiment, step 3 may refer to" calculating product return values of a plurality of pairing combinations respectively based on a preset return function, where a product vector or an attribute vector corresponding to a pairing combination corresponding to the product return value with the largest value is a product vector or an attribute vector matching the target user vector. When the attribute vector is matched with the target user vector, carrying out preference inquiry of corresponding attributes on the target user in the next round of conversation; and when the product vector is matched with the target user vector, recommending the product to the target user in the next conversation. "step 4 may refer to" when the current number of dialog turns is higher than the preset number of dialog turns threshold, the dialog is ended. "step 5 may refer to the above embodiment for" when the feedback is positive feedback and the response is attribute vector, add the attribute vector to the target user interest pool, filter the products in the candidate pool that do not include the attribute, update the target user vector according to the feedback, and perform vector matching and query again; when the feedback is positive feedback and the response is a product vector, recommending a product corresponding to the product vector to the target user, and ending; when the feedback is negative feedback, removing the corresponding attribute vector or product vector in the candidate pool, updating the target user vector according to the feedback, and performing vector matching and querying again. "is described.
Fig. 2 is a schematic structural diagram of a dialog-based product recommendation apparatus provided in an embodiment of the present disclosure, including:
a vector acquisition module 201, which acquires a target user vector;
the vector matching module 202 is used for performing vector matching on the target user vector through the attribute vector in the target user interest pool, and determining a product vector or an attribute vector matched with the target user vector in the candidate pool;
the statement confirmation module 203 is used for determining an inquiry statement when the target user carries out conversation according to the product vector or the attribute vector matched with the target user vector and receiving feedback responding to the inquiry statement;
the first execution module 204 is used for adding the attribute vector to a target user interest pool when the feedback is positive feedback and the response is the attribute vector, filtering product vectors which do not comprise the attribute vector in a candidate pool, updating the target user vector according to the feedback, and executing vector matching and query again;
the second execution module 205, recommending a product corresponding to the product vector to the target user when the feedback is positive feedback and the response is a product vector, and ending;
and the third execution module 206, when the feedback is negative feedback, removes the corresponding attribute vector or product vector in the candidate pool, updates the target user vector according to the feedback, and executes vector matching and query again.
Further, the vector obtaining module 201 is configured to obtain historical user interaction data; based on the characteristic relevance of the historical user interaction data, a historical user vector, a product vector and an attribute vector are respectively constructed, wherein the historical user vector is a target user vector for carrying out conversation with a target user for the first time.
Further, the vector matching module 202 is configured to pair the target user vector and the product vector and/or the attribute vector in the candidate pool in sequence through the attribute vector in the target user interest pool to obtain a plurality of pairing combinations.
Further, the vector matching module 202 is configured to calculate product return values of a plurality of pairing combinations respectively based on a preset return function, where a product vector or an attribute vector corresponding to a pairing combination corresponding to the product return value with a largest value is a product vector or an attribute vector matched with the target user vector.
Further, the statement confirming module 203 is configured to perform preference query on the corresponding attribute for the target user in a next session when the attribute vector is matched with the target user vector; and when the product vector is matched with the target user vector, recommending the product to the target user in the next conversation.
Further, the first executing module 204 is configured to initialize a target user vector that follows a multidimensional gaussian distribution, where the target user vector includes a new user vector and/or an inactive user vector; acquiring the current number of conversation turns of the target user; when the current number of turns is lower than a preset number of turns threshold, updating parameters which are distributed from a multidimensional Gaussian in a target user vector by responding to feedback of the query statement to obtain an updated target user vector; and when the current conversation turn number is higher than a preset conversation turn number threshold value, ending the conversation.
Further, a third executing module 206, configured to initialize a target user vector that is subject to multidimensional gaussian distribution, where the target user vector includes a new user vector and/or an inactive user vector; acquiring the current number of conversation turns of the target user; when the current number of turns is lower than a preset number of turns threshold, updating parameters which are distributed from a multidimensional Gaussian in a target user vector by responding to feedback of the query statement to obtain an updated target user vector; and when the current conversation turn number is higher than a preset conversation turn number threshold value, ending the conversation.
The functions of the apparatus in the embodiment of the present invention have been described in the above method embodiments, so that reference may be made to the related descriptions in the foregoing embodiments for details that are not described in the present embodiment, and further details are not described herein.
Based on the same inventive concept, the embodiment of the specification further provides the electronic equipment.
In the following, embodiments of the electronic device of the present invention are described, which may be regarded as specific physical implementations for the above-described embodiments of the method and apparatus of the present invention. Details described in the embodiments of the electronic device of the invention should be considered supplementary to the embodiments of the method or apparatus described above; for details which are not disclosed in embodiments of the electronic device of the invention, reference may be made to the above-described embodiments of the method or the apparatus.
Fig. 3 is a schematic structural diagram of an electronic device provided in an embodiment of the present specification. An electronic device 300 according to this embodiment of the invention is described below with reference to fig. 3. The electronic device 300 shown in fig. 3 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present invention.
As shown in fig. 3, electronic device 300 is embodied in the form of a general purpose computing device. The components of electronic device 300 may include, but are not limited to: at least one processing unit 310, at least one memory unit 320, a bus 330 connecting different device components (including the memory unit 320 and the processing unit 310), a display unit 340, and the like.
Wherein the storage unit stores program code executable by the processing unit 310 to cause the processing unit 310 to perform the steps according to various exemplary embodiments of the present invention described in the above-mentioned processing method section of the present specification. For example, the processing unit 310 may perform the steps as shown in fig. 1.
The storage unit 320 may include readable media in the form of volatile storage units, such as a random access memory unit (RAM)3201 and/or a cache storage unit 3202, and may further include a read only memory unit (ROM) 3203.
The storage unit 320 may also include a program/utility 3204 having a set (at least one) of program modules 3205, such program modules 3205 including, but not limited to: an operating device, 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.
The electronic device 300 may also communicate with one or more external devices 400 (e.g., keyboard, pointing device, bluetooth device, etc.), with one or more devices that enable a user to interact with the electronic device 300, and/or with any devices (e.g., router, modem, etc.) that enable the electronic device 300 to communicate with one or more other computing devices. Such communication may occur via an input/output (I/O) interface 350. Also, the electronic device 300 may 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) via the network adapter 360. Network adapter 360 may communicate with other modules of electronic device 300 via bus 330. It should be appreciated that although not shown in FIG. 3, other hardware and/or software modules may be used in conjunction with electronic device 300, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID devices, tape drives, and data backup storage devices, to name a few.
Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments of the present invention described herein may be implemented by software, or by software in combination with necessary hardware. Therefore, the technical solution according to the embodiment of the present invention can be embodied in the form of a software product, which can be stored in a computer-readable storage medium (which can be a CD-ROM, a usb disk, a removable hard disk, etc.) or on a network, and includes several instructions to make a computing device (which can be a personal computer, a server, or a network device, etc.) execute the above-mentioned method according to the present invention. The computer program, when executed by a data processing apparatus, enables the computer readable medium to implement the above-described method of the invention, namely: such as the method shown in fig. 1.
Fig. 4 is a schematic diagram of a computer-readable medium provided in an embodiment of the present disclosure.
A computer program implementing the method shown in fig. 1 may be stored on one or more computer readable media. The computer readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor device, apparatus, or device, or a combination of any 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.
The computer readable storage 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 storage 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 apparatus, device, or apparatus. Program code embodied on a readable storage 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 aspects of the present invention 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).
In summary, the invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that some or all of the functionality of some or all of the components in embodiments in accordance with the invention may be implemented in practice using a general purpose data processing device such as a microprocessor or a Digital Signal Processor (DSP). The present invention may also be embodied as apparatus or device programs (e.g., computer programs and computer program products) for performing a portion or all of the methods described herein. Such programs implementing the present invention may be stored on computer-readable media or may be in the form of one or more signals. Such a signal may be downloaded from an internet website or provided on a carrier signal or in any other form.
While the foregoing embodiments have described the objects, aspects and advantages of the present invention in further detail, it should be understood that the present invention is not inherently related to any particular computer, virtual machine or electronic device, and various general-purpose machines may be used to implement the present invention. The invention is not to be considered as limited to the specific embodiments thereof, but is to be understood as being modified in all respects, all changes and equivalents that come within the spirit and scope of the invention.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments.
The above description is only an example of the present application and is not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims (10)
1. A dialog-based product recommendation method, comprising:
acquiring a target user vector;
carrying out vector matching on the target user vector through the attribute vector in the target user interest pool, and determining a product vector or an attribute vector matched with the target user vector in a candidate pool;
determining an inquiry statement when the target user carries out conversation according to the product vector or the attribute vector matched with the target user vector, and receiving feedback responding to the inquiry statement;
when the feedback is positive feedback and the response is attribute vector, adding the attribute vector to a target user interest pool, filtering product vectors which do not comprise the attribute vector in a candidate pool, updating the target user vector according to the feedback, and executing vector matching and inquiring again;
when the feedback is positive feedback and the response is a product vector, recommending a product corresponding to the product vector to the target user, and ending;
when the feedback is negative feedback, removing the corresponding attribute vector or product vector in the candidate pool, updating the target user vector according to the feedback, and performing vector matching and querying again.
2. The dialog-based product recommendation method of claim 1, wherein the obtaining a target user vector comprises:
acquiring historical user interaction data;
based on the characteristic relevance of the historical user interaction data, a historical user vector, a product vector and an attribute vector are respectively constructed, wherein the historical user vector is a target user vector for carrying out conversation with a target user for the first time.
3. The dialog-based product recommendation method of claim 1 or 2, wherein the vector matching of the target user vector through the attribute vectors in the target user interest pool comprises:
and pairing the target user vector and the product vector and/or the attribute vector in the candidate pool in sequence through the attribute vector in the target user interest pool to obtain a plurality of pairing combinations.
4. The dialog-based product recommendation method of any one of claims 1-3 wherein determining a product vector or an attribute vector in the candidate pool that matches the target user vector comprises:
and respectively calculating product return values of a plurality of pairing combinations based on a preset return function, wherein a product vector or an attribute vector corresponding to the pairing combination corresponding to the product return value with the largest numerical value is a product vector or an attribute vector matched with the target user vector.
5. The dialog-based product recommendation method of any one of claims 1-4 wherein determining the query statement in dialog with the target user comprises:
when the attribute vector is matched with the target user vector, carrying out preference inquiry of corresponding attributes on the target user in the next round of conversation;
and when the product vector is matched with the target user vector, recommending the product to the target user in the next conversation.
6. The dialog-based product recommendation method of any one of claims 1-5 wherein said updating a target user vector based on said feedback comprises:
initializing a target user vector obeying multi-dimensional Gaussian distribution, wherein the target user vector comprises a new user vector and/or an inactive user vector;
acquiring the current number of conversation turns of the target user;
when the current number of turns is lower than a preset number of turns threshold, updating parameters which are distributed from a multidimensional Gaussian in a target user vector by responding to feedback of the query statement to obtain an updated target user vector;
and when the current conversation turn number is higher than a preset conversation turn number threshold value, ending the conversation.
7. A dialog-based product recommendation device, comprising:
the vector acquisition module is used for acquiring a target user vector;
the vector matching module is used for carrying out vector matching on the target user vector through the attribute vector in the target user interest pool and determining a product vector or an attribute vector matched with the target user vector in the candidate pool;
the statement confirmation module is used for determining an inquiry statement when the target user carries out conversation according to the product vector or the attribute vector matched with the target user vector and receiving feedback responding to the inquiry statement;
the first execution module is used for adding the attribute vector to a target user interest pool when the feedback is positive feedback and the response is the attribute vector, filtering product vectors which do not comprise the attribute vector in a candidate pool, updating the target user vector according to the feedback, and executing vector matching and inquiring again;
the second execution module recommends a product corresponding to the product vector to the target user when the feedback is positive feedback and the response is the product vector, and the operation is finished;
and the third execution module is used for removing the corresponding attribute vector or product vector in the candidate pool when the feedback is negative feedback, updating the target user vector according to the feedback, and executing vector matching and inquiry again.
8. The dialog-based product recommendation device of claim 7 wherein determining a product vector or an attribute vector in the candidate pool that matches the target user vector comprises:
and respectively calculating product return values of a plurality of pairing combinations based on a preset return function, wherein a product vector or an attribute vector corresponding to the pairing combination corresponding to the product return value with the largest numerical value is a product vector or an attribute vector matched with the target user vector.
9. An electronic device, wherein the electronic device comprises:
a processor and a memory storing computer-executable instructions that, when executed, cause the processor to perform the method of any of claims 1-6.
10. A computer readable storage medium, wherein the computer readable storage medium stores one or more programs which, when executed by a processor, implement the method of any of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110707740.9A CN113487379B (en) | 2021-06-24 | 2021-06-24 | Product recommendation method and device based on conversation mode and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110707740.9A CN113487379B (en) | 2021-06-24 | 2021-06-24 | Product recommendation method and device based on conversation mode and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113487379A true CN113487379A (en) | 2021-10-08 |
| CN113487379B CN113487379B (en) | 2023-01-13 |
Family
ID=77936263
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110707740.9A Active CN113487379B (en) | 2021-06-24 | 2021-06-24 | Product recommendation method and device based on conversation mode and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113487379B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107729469A (en) * | 2017-10-12 | 2018-02-23 | 北京小度信息科技有限公司 | Usage mining method, apparatus, electronic equipment and computer-readable recording medium |
| US20190034810A1 (en) * | 2017-07-31 | 2019-01-31 | Boe Technology Group Co., Ltd. | Data processing method, medical term processing system and medical diagnostic system |
| CN109460512A (en) * | 2018-10-25 | 2019-03-12 | 腾讯科技（北京）有限公司 | Recommendation information processing method, device, equipment and storage medium |
| CN110599307A (en) * | 2019-09-17 | 2019-12-20 | 北京三快在线科技有限公司 | Commodity recommendation method and device |
| CN112883170A (en) * | 2021-01-20 | 2021-06-01 | 中国人民大学 | User feedback guided self-adaptive conversation recommendation method and system |
| CN112925892A (en) * | 2021-03-23 | 2021-06-08 | 苏州大学 | Conversation recommendation method and device, electronic equipment and storage medium |
-
2021
- 2021-06-24 CN CN202110707740.9A patent/CN113487379B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190034810A1 (en) * | 2017-07-31 | 2019-01-31 | Boe Technology Group Co., Ltd. | Data processing method, medical term processing system and medical diagnostic system |
| CN107729469A (en) * | 2017-10-12 | 2018-02-23 | 北京小度信息科技有限公司 | Usage mining method, apparatus, electronic equipment and computer-readable recording medium |
| CN109460512A (en) * | 2018-10-25 | 2019-03-12 | 腾讯科技（北京）有限公司 | Recommendation information processing method, device, equipment and storage medium |
| CN110599307A (en) * | 2019-09-17 | 2019-12-20 | 北京三快在线科技有限公司 | Commodity recommendation method and device |
| CN112883170A (en) * | 2021-01-20 | 2021-06-01 | 中国人民大学 | User feedback guided self-adaptive conversation recommendation method and system |
| CN112925892A (en) * | 2021-03-23 | 2021-06-08 | 苏州大学 | Conversation recommendation method and device, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113487379B (en) | 2023-01-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11068285B2 (en) | Machine-learning models applied to interaction data for determining interaction goals and facilitating experience-based modifications to interface elements in online environments | |
| US20150058331A1 (en) | Search result ranking using machine learning | |
| TW201822104A (en) | Intelligent recommendation method and system | |
| US9141671B2 (en) | System and method for presenting filter options to a user based on ongoing monitoring of filter selections | |
| CN112199623A (en) | Script execution method and device, electronic equipment and storage medium | |
| MX2013013451A (en) | System and method for classifying relevant competitors. | |
| JP5801257B2 (en) | Product diversification recommendation device, method and program | |
| CN111209477A (en) | Information recommendation method and device, electronic equipment and storage medium | |
| CN102289759A (en) | Item recommendation system, item recommendation method and program | |
| CA2953775A1 (en) | Systems and methods for serving product recommendations | |
| KR102307517B1 (en) | Method and apparatus of recommending goods based on network | |
| CN111275205A (en) | Virtual sample generation method, terminal device and storage medium | |
| US20200356360A1 (en) | Automated software selection using matrix factorization | |
| KR20200140588A (en) | System and method for providing image-based service to sell and buy product | |
| CN114817746A (en) | Insurance product recommendation method, device, equipment and storage medium | |
| US11461824B2 (en) | Systems and methods of product recommendation and integrated language modelling | |
| CN112950320A (en) | Automatic commodity online and offline method and device based on machine self-learning and electronic equipment | |
| CN111461757A (en) | Information processing method and device, computer storage medium and electronic equipment | |
| WO2012034606A2 (en) | Multiverse recommendation method for context-aware collaborative filtering | |
| CN110598094A (en) | Shopping recommendation method based on matrix completion, electronic device and storage medium | |
| JP5945206B2 (en) | Product recommendation device, method and program | |
| CN110288444A (en) | Realize the method and system of user's associated recommendation | |
| US20120109783A1 (en) | Product information search | |
| CN113487379B (en) | Product recommendation method and device based on conversation mode and electronic equipment | |
| KR20230009437A (en) | User search category predictor |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP01 | Change in the name or title of a patent holder | ||
| CP01 | Change in the name or title of a patent holder |
Address after: 200333 room 1027, 10th floor, 89 Yunling East Road, Putuo District, Shanghai Patentee after: Qifu Shuke (Shanghai) Technology Co.,Ltd. Address before: 200333 room 1027, 10th floor, 89 Yunling East Road, Putuo District, Shanghai Patentee before: Shanghai Qifu Information Technology Co.,Ltd. |