A consultation processing method and apparatus

Abstract

The application discloses a kind of consultation processing method and device, it is related to computer technical field.The specific embodiment of the method includes: receiving user consultation behavior, capture user access page behavior track information and access frequency information, to locate the node where user is currently in in behavior track diagram;The structure of behavior track diagram is disassembled, to determine the level where the node is located, retrieve the historical intent corresponding to other nodes in each level except the node;The effective value of each historical intent is calculated, and the target historical intent is screened out according to level and effective value size, and the corresponding operation is carried out using the decision scheme corresponding to the target historical intent.This embodiment carries out intelligent audit by the historical intent in the consultation information and behavior track diagram of user, to improve the identification rate of user real consultation intent.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, in particular to a consultation processing method and device. BACKGROUND

[0002] In the e-commerce scenario, the number of after-sales service orders is large, and the proportion of intelligent auditing is relatively small compared with manual customer service auditing. Therefore, it is necessary to quickly expand the proportion to relieve the pressure of manual customer service and improve the intelligent progress of the entire after-sales service. Currently, there are two ways:

[0003] 1) Statistical-based method: relying on user historical consultation information, recording user historical consultation intent and solution, if the same consultation is encountered again, the relevant historical information can be directly obtained and the same solution is recommended to the user.

[0004] 2) Semantic rule-based method: according to the user's consultation scenario, algorithm analysis is performed, the user's intent is located according to the analysis result, and the corresponding solution is pushed, wherein the intent recognition technology and manual sorting technology are mainly used.

[0005] In the process of implementing the present application, the inventors found that the existing technology at least has the following problems: the first method has dispersion, which needs to separate the user, and the intent recognition and recommendation requirements need a lot of labeling and implementation, and the heuristic between different users cannot be used; the second method has redundancy, when a new activity occurs, the relevant content needs to be added according to the user's intent, and the historical model cannot add new intent because it has not used new data, so there is no corresponding solution. SUMMARY

[0006] Therefore, the present application provides a consultation processing method and device, which can at least solve the dispersion and redundancy of the existing consultation processing scheme.

[0007] To achieve the above-mentioned purpose, according to one aspect of the present application, a consultation processing method is provided, comprising:

[0008] Receiving user consultation behavior, capturing user access page behavior track information and access frequency information to locate the node where the user is currently located in the behavior track graph; wherein the behavior track graph is composed of multiple nodes connected by lines;

[0009] The structure of the behavior track graph is disassembled to determine the level of the node, and the historical intent corresponding to other nodes in each level except the node is retrieved;

[0010] The effective value of each historical intent is calculated, the target historical intent is selected according to the level and the effective value, and the corresponding operation is performed using the decision scheme corresponding to the target historical intent.

[0011] Optionally, the behavior track information and the access frequency information of the user accessing the page are captured, including:

[0012] From the moment when the user initiates the consultation, the chat information is captured for rough positioning to find the user basic information; wherein the chat information includes the consultation content and the reply content;

[0013] Based on the user basic information, the behavior track information and the access frequency information of the user accessing the page are captured; wherein the behavior track information is generated by capturing the track of the user browsing different pages, and the access frequency information is generated by counting the number of times of the user accessing each page.

[0014] Optionally, the node where the user currently stays in the behavior track graph is located, including:

[0015] All historical behavior tracks corresponding to the username of the user within a preset time period from the current time are queried, and then it is queried whether the all historical behavior tracks exist in the behavior track graph;

[0016] If the all historical behavior tracks exist, the node where the user currently stays in the behavior track graph is located, otherwise the all historical behavior tracks are added to the behavior track graph to obtain a new behavior track graph, and a new node is inserted; and

[0017] It is detected whether the access frequency information is contained under the node / new node, if not, the access frequency information is added under the node / new node.

[0018] Optionally, the node / new node is highlighted in the behavior track graph.

[0019] Optionally, the structure of the behavior track graph is disassembled to determine the level where the node stays, including:

[0020] The connection nodes connected with the node in the behavior track graph are determined, for any node in the connection nodes, the any node is classified into a corresponding number of Ks layers according to the number of lines connected with the any node;

[0021] For the node, after removing the connection nodes that have been classified into the Ks=1 layer, it is judged whether the remaining connection nodes are in the same Ks layer, if yes, the node is classified into a corresponding number of Ks layers according to the number of the remaining connection nodes; or

[0022] If not, the connection nodes that have been classified into the Ks=2 layer are continuously removed, it is judged whether the remaining connection nodes are in the same Ks layer, if yes, the node is classified into a corresponding number of Ks layers according to the number of the remaining connection nodes;

[0023] Repeat the above process to attribute each connection node and each node to a corresponding Ks layer.

[0024] Optionally, after the connection node connected with the node in the determined behavior trajectory graph, further comprising:

[0025] For any node in the connection node, count the number of associations between the remaining nodes in the behavior trajectory graph except the any node;

[0026] The ratio of the number of associations to the number of other remaining nodes is taken as the degree of the any node, if the degree is 1, the any node is removed, otherwise it is retained;

[0027] Repeat the above degree calculation and processing process until there is no connection node with a degree of 1.

[0028] Optionally, before the structure of the behavior trajectory graph is disassembled, further comprising: merging nodes with the same name in the behavior trajectory graph.

[0029] Optionally, the retrieving the historical intention corresponding to each node in the hierarchy except the node includes: retrieving the historical user corresponding to the other node, and obtaining the corresponding historical intention based on the historical behavior trajectory of each historical user;

[0030] The calculation of the effective value of each historical intention includes: calculating the effective value of each historical intention using a graph classification method, and filtering the historical intention with an effective value lower than the effective value threshold.

[0031] Optionally, the using the decision scheme corresponding to the target historical intention to perform the corresponding operation further includes:

[0032] Sending the target historical intention to the user, if the user inputted an adoption message, using the decision scheme corresponding to the target historical intention to perform the corresponding operation; or

[0033] If no adoption message is received, sending the intention query information to the user, then receiving the real intention inputted by the user, finding a historical intention with the highest similarity / association degree with the real intention from the historical intentions, and then using the decision scheme corresponding to the one historical intention to perform the corresponding operation.

[0034] Optionally, the target historical intention and the decision scheme are injected into the node.

[0035] To achieve the above object, according to another aspect of the embodiment of the present application, a consultation processing device is provided, comprising:

[0036] A positioning module is configured to receive user consultation behavior, capture behavior track information and access frequency information of a user accessing a page, and locate a node where the user is currently located in a behavior track graph; wherein the behavior track graph is formed by connecting a plurality of nodes;

[0037] A disassembling module is configured to disassemble a structure of the behavior track graph to determine a level where the node is located, and retrieve historical intentions corresponding to other nodes in each level except the node;

[0038] A screening module is configured to calculate an effective value of each historical intention, screen a target historical intention according to the level and the effective value, and perform a corresponding operation using a decision scheme corresponding to the target historical intention.

[0039] Optionally, the positioning module is configured to:

[0040] Starting from a time when the user initiates consultation, chat information is captured for rough positioning to find user basic information; wherein the chat information includes consultation content and reply content;

[0041] Based on the user basic information, behavior track information and access frequency information of the user accessing the page are captured; wherein the behavior track information is generated by capturing a track of the user browsing different pages, and the access frequency information is generated by counting a number of times the user accesses each page.

[0042] Optionally, the positioning module is configured to:

[0043] All historical behavior tracks corresponding to a username of the user within a current preset time period are queried, and then it is determined whether the all historical behavior tracks exist in the behavior track graph;

[0044] If the all historical behavior tracks exist, a node where the user is currently located in the behavior track graph is located, otherwise the all historical behavior tracks are added to the behavior track graph to obtain a new behavior track graph, and a new node is inserted; and

[0045] It is detected whether the access frequency information is included under the node / the new node, and if not, the access frequency information is added under the node / the new node.

[0046] Optionally, the node / the new node is highlighted in the behavior track graph.

[0047] Optionally, the disassembling module is configured to:

[0048] A connection node having a connection relationship with the node in the behavior track graph is determined, and for any node in the connection node, the any node is attributed to a corresponding number of Ks layers according to a number of connection lines of the any node.

[0049] For the node, after removing the connection nodes that have been attributed to the Ks=1 layer, it is judged whether the remaining connection nodes are all in the same Ks layer, if yes, the node is attributed to the corresponding number of Ks layers according to the number of the remaining connection nodes; or

[0050] If not, the connection nodes that have been attributed to the Ks=2 layer are continuously removed, it is judged whether the remaining connection nodes are all in the same Ks layer, if yes, the node is attributed to the corresponding number of Ks layers according to the number of the remaining connection nodes;

[0051] The above process is repeated to attribute each connection node and each node to the corresponding Ks layer.

[0052] Optionally, the disassembling module is further configured to:

[0053] For any node in the connection nodes, the number of associations between the any node and the other remaining nodes in the behavior trajectory graph is counted;

[0054] The ratio of the number of associations to the number of the other remaining nodes is taken as the degree of the any node, if the degree is 1, the any node is removed, otherwise, the any node is retained;

[0055] The degree calculation and processing process is repeated until there is no connection node with a degree of 1.

[0056] Optionally, the disassembling module is further configured to: merge the nodes with the same name in the behavior trajectory graph.

[0057] Optionally, the disassembling module is configured to: retrieve the historical users corresponding to the other nodes, and obtain the corresponding historical intentions based on the historical behavior trajectories of each historical user.

[0058] The screening module is configured to: calculate the effective value of each historical intention by using a graph classification method, and filter the historical intentions with an effective value lower than an effective value threshold.

[0059] Optionally, the screening module is further configured to:

[0060] The target historical intention is sent to the user, if an adoption message input by the user is received, a corresponding operation is performed by using a decision scheme corresponding to the target historical intention; or

[0061] If no adoption message is received, intention inquiry information is sent to the user, and then a real intention input by the user is received, one historical intention with the highest similarity / association degree to the real intention is found from the historical intentions, and then a corresponding operation is performed by using a decision scheme corresponding to the one historical intention.

[0062] Optionally, the target historical intention and the decision scheme are injected into the node.

[0063] To achieve the above object, according to another aspect of the embodiments of the present application, there is provided an electronic device for consultation processing.

[0064] The electronic device according to the embodiments of the present application comprises one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the consultation processing method as described above.

[0065] To achieve the above object, according to another aspect of the embodiments of the present application, there is provided a computer readable medium having stored thereon a computer program, which when executed by a processor implements the consultation processing method as described above.

[0066] According to the scheme provided by the present application, one embodiment of the above application has the following advantages or beneficial effects: 1) For the dispersion problem of the system, a user-associated history access complementary architecture model is provided, which can automatically learn the possible similar needs between users and associate them, and use big data mechanism for comparative learning, and the similar intentions between users can also be directly recommended for solution.

[0067] 2) For the redundancy problem of the system, a new intention fission and automatic solution generation solution is provided, which can effectively calculate the user's intention and the existing intention through the intention calculation in the intention recognition process, if it is found that the existing intention cannot cover all possible intentions, then it is induced into a new intention way, and then through the user's behavior analysis and the related results in the user group, the possible solution is given, and it is split into related possible solutions.

[0068] The further effects of the above-mentioned non-conventional optional mode will be described in the following combined with the specific embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0069] The accompanying drawings are used to better understand the present application, and do not constitute an improper limitation on the present application. Among them:

[0070] Figure 1 is the main flowchart of a consultation processing method according to an embodiment of the present application;

[0071] Figure 2 is the flowchart of an optional consultation processing method according to an embodiment of the present application;

[0072] Figure 3 is the flowchart of another optional consultation processing method according to an embodiment of the present application;

[0073] Figure 4 is a flowchart of yet another optional consultation processing method according to an embodiment of the present application;

[0074] Fig. 5(a)-(d) is a flowchart of constructing a KS layer in detail;

[0075] Figure 6 is a main module diagram of a consultation processing device according to an embodiment of the present application;

[0076] Figure 7 is an exemplary system architecture diagram to which embodiments of the present application can be applied;

[0077] Figure 8 is a structural diagram of a computer system of a mobile device or server suitable for implementing embodiments of the present application. DETAILED DESCRIPTION

[0078] Exemplary embodiments of the present application are described below with reference to the accompanying drawings, which include various details of the embodiments of the present application to assist in understanding them. These should be considered in their context only. Thus, those of ordinary skill in the art will recognize the various changes and modifications of the embodiments described herein, without departing from the scope and spirit of the present application. Also, descriptions of well-known functions and structures are omitted in the following description for clarity and conciseness.

[0079] Referring to Figure 1 , a main flowchart of a consultation processing method provided by an embodiment of the present application is shown, including the following steps:

[0080] S101: receiving a user consultation behavior, capturing behavior trajectory information and access frequency information of a user accessing a page, to locate a node where the user is currently located in a behavior trajectory graph; wherein the behavior trajectory graph is composed by connecting a plurality of nodes;

[0081] S102: disassembling the structure of the behavior trajectory graph to determine the level of the node, and retrieving historical intentions corresponding to other nodes in each level except the node;

[0082] S103: calculating the effective value of each historical intention, screening out a target historical intention according to the level and the effective value size, and using a decision scheme corresponding to the target historical intention to perform corresponding operations.

[0083] In the above embodiment, for step S101, the present scheme is mainly applicable to an e-commerce scenario. At present, there are a large number of after-sales service orders, but there are fewer after-sales manual customer service. If intelligent operation is applied to after-sales intelligent auditing, a large amount of manpower can be saved, thereby saving labor costs.

[0084] In the process of chatting with customer service or a merchant, a user usually contains multiple consultation intentions, such as an intention of recommending goods, an intention of processing after-sales quality problems, an intention of processing invoices and price protection, an intention of consulting detailed information of goods, and the like. Therefore, starting from the moment when the user initiates a consultation behavior, chat information of the user and the customer service or the merchant is captured for rough positioning to find basic information of the user; the chat information includes content consulted by the user and content replied by the customer service or the merchant, and the basic information at least includes a user name and a PIN (Personal Identification Number).

[0085] Based on the basic information of the user, important behaviors of the user in accessing pages are captured, and the behaviors are obtained from two dimensions: 1) an order in which the user browses different pages, so as to generate behavior track information; and 2) a number of times of browsing, for example, a number of times of accessing a certain page by the user, as access frequency information.

[0086] A specific process of positioning a node currently located by the user in the behavior track graph is shown in Figure 2

[0087] S201: query all historical behavior tracks corresponding to the user name within a preset time period from a current time, and then query whether the all historical behavior tracks exist in a behavior track graph;

[0088] S202: if the all historical behavior tracks exist, a node currently located by the user is positioned in the behavior track graph;

[0089] S203: otherwise, the all historical behavior tracks are added to the behavior track graph to obtain a new behavior track graph, and a new node is inserted;

[0090] S204: it is detected whether the access frequency information is included under the node / the new node, and if not, the access frequency information is added under the node / the new node.

[0091] For steps S201-S203, a complete service in an e-commerce scenario includes multiple pages, such as a goods detail page, a shopping cart page, an order page, an order page, a logistics page, an invoice page, a return and exchange page, a refund page, a complaint page, a dispute page, and a compensation page, each page corresponds to a node, and a complete behavior track graph is composed of multiple nodes connected by lines. A user may access multiple pages since registration, in order to narrow the processing range, only historical behavior tracks within a current half hour (only an example, the value can be adjusted) are found according to the user name, and the number of the historical behavior tracks may be more than one.

[0092] ​If the behavior trajectory graph does not contain the user's historical behavior trajectory this time, such as visiting a new activity page, the user's historical behavior trajectory this time is added to the behavior trajectory graph for updating to obtain a new behavior trajectory graph. In addition, since the node corresponding to the new activity page does not exist in the original behavior trajectory graph, a new node needs to be inserted, and the new node is taken as the node where the user is currently located. Otherwise, the node position where the user is currently located is directly located in the original behavior trajectory graph.

[0093] It should be noted that the number of determined nodes is usually greater than one, for example, the user clicks the shopping cart interface, selects one of the items and completes the order operation, and then clicks the refund operation, at this time the number of nodes is 3, which correspond to the shopping cart page node, the order page node and the refund page node respectively.

[0094] For step S204, it is continued to detect whether the access frequency information of the user is located under the node / new node, and if not, the access frequency information is added to the determined node / new node.

[0095] After inserting the user behavior trajectory information and the access frequency information, the nodes with the same name in the behavior trajectory graph are merged. In addition, the behavior trajectory information of the user, the access frequency information, and the node / new node in the behavior trajectory graph can be highlighted for subsequent management.

[0096] For step S102, this step mainly describes how to construct the Ks layer. First, the nodes in the behavior trajectory graph need to be filtered, which is specifically described in Figure 3 as shown in the following figure:

[0097] S301: determining a connection node connected with the node in the behavior trajectory graph;

[0098] S302: for any node in the connection node, counting the number of associations between the other remaining nodes in the behavior trajectory graph except the any node;

[0099] S303: taking the ratio of the association number to the number of the other remaining nodes as the degree of the any node, if the degree is 1, the any node is removed, otherwise it is reserved;

[0100] S304: repeating the degree calculation and processing process until there is no connection node with a degree of 1.

[0101] The structure of the behavior trajectory graph is decomposed by using the degree expansion method. First, the node with a degree of 1 is removed from the highlighted node. The calculation method is as follows:

[0102]

[0103] Wherein, the node i is any node which has connection relationship with the highlighted node, the numerator represents the number of association between the node i and other (g-1) nodes in the behavior trajectory graph, and the denominator represents the maximum connection number (g-1). G' D (N i ) The proportion range is 0.0-1.0, 0.0 represents no contact with each of the other nodes, 1.0 represents direct contact with each of the other nodes, and the node i is removed with the degree of 1. The above operation is repeated until the degree of any node which has connection relationship with the highlighted node is not 1.

[0104] After filtering the nodes, the Ks layer is constructed (the idea that the node importance depends on its position in the whole network, and the node importance ranking index (k-shell) is obtained by using K-core decomposition, see Figure 4 as shown:

[0105] S401: For any node in the connection node, according to the number of connection lines of the any node, the any node is attributed to the corresponding number of Ks layers;

[0106] S402: For the node, after removing the connection nodes which have been attributed to the Ks=1 layer, it is judged whether the remaining connection nodes are in the same Ks layer, if yes, according to the number of the remaining connection nodes, the node is attributed to the corresponding number of Ks layers;

[0107] S403: If not, continue to remove the connection nodes which have been attributed to the Ks=2 layer, judge whether the remaining connection nodes are in the same Ks layer, if yes, according to the number of the remaining connection nodes, the node is attributed to the corresponding number of Ks layers;

[0108] S404: Repeat the above process to attribute each connection node and each node to the corresponding Ks layer.

[0109] Referring to Fig. 5(a), the black node is the highlighted node, and the rest are nodes which have connection relationship with the highlighted node. The two white nodes in the upper right corner, the number of connection lines of which with other nodes (here, the highlighted node) is only 1, so the two nodes are attributed to the Ks=1 layer. For the two nodes in the upper left corner, the number of connection lines of which with other nodes (here, including the highlighted node and the white node) is 2, so the two nodes are attributed to the Ks=2 layer. For the three nodes in the middle, the number of connection lines of which with other nodes is 3, so the three nodes are attributed to the Ks=3 layer.

[0110] For the highlighted node in the upper right corner, after removing the connection nodes which have been attributed to the Ks=1 layer, the number of the remaining nodes which have connection relationship with it is only 1, and the node is in the Ks=3 layer, so the highlighted node is also attributed to the Ks=1 layer.

[0111] For the highlighted node in the middle, after removing the connecting nodes already assigned to layer Ks=1, some of the remaining connected nodes are located at layer Ks=2, while others are located at layer Ks=3. Therefore, the aforementioned method cannot be used, and further judgment is required. Here, we continue removing the connecting nodes already assigned to layer Ks=2. All remaining connecting nodes are located at layer Ks=3, and the total number is 3. Therefore, this highlighted node is assigned to layer Ks=3.

[0112] In practice, the number of Ks layers may be greater than 3. The above judgment can be repeated until each node is assigned to the corresponding Ks layer. This can be repeated to obtain Ks=4, Ks=5, etc. See Figures 5(b) to (c).

[0113] After determining the Ks level of each highlighted node, the historical users corresponding to each node in each Ks level (excluding the highlighted node) are obtained. Based on the historical behavior trajectory of each historical user, the corresponding historical intent is obtained. The historical intents are then sorted in descending order of level to obtain the historical intent sequence. At this point, the intent of the highlighted node has not yet been determined.

[0114] For step S103, a graph classification method is used to review each historical intent in the historical intent sequence sorted by Ks, using the formula:

[0115]

[0116]

[0117] Here, f1 and f2 represent two transformation functions, such as linear functions and quadratic functions, which can take various forms such as MLP (Multi-Layer Perceptron), CNN (Convolutional Neural Network), and GRU (Gate Recurrent Unit). σ represents sigmoid activation.

[0118] For formula (1), again taking Figure 5(a) as an example, there are a total of 9 nodes, each numbered 1 to 9, as shown in Figure 5(d). Setting The vector format is [x,x,x,x,x,x,x,x,x] (x is 0 or 1). Taking node 1 as an example, it is only connected to node 2. As for node 7, it is connected to nodes 4, 5, 6, 8, and 9. Each node Inputting the data into formula (1) yields the effective value h for each historical intent. v .

[0119] For formula (2), first sum up h v of all nodes, and then take the average to get the normalized result. For Maxpooling(h1...h v ), it is to extract the maximum value in each dimension from the 9 vectors h1...h9 vertically to get a new vector of the maximum pooling, which is still 1x9 dimension. The two are spliced to get the hidden layer vector (effective value threshold) h θ of all historical intentions. v The h v is audited to eliminate one or more historical intentions with h θ .

[0120] The above operation mode uses the degree measurement and the calculation method of the graph network to effectively grasp the real message concerned by the user in the sorting and recommendation of the intention, and automatically recommends the related goods after dividing the real message. The self-splitting mode of the graph is used, and the efficiency in mining new nodes is greatly improved.

[0121] In actual operation, the number of intentions in the library is large, such as 3000, so even if the above operation is performed, multiple historical intentions are still obtained, and if these historical intentions are sent to the user together, it may produce a chaotic visual effect, affecting the user experience. Preferably, one historical intention with the largest effective value is extracted from the highest Ks level as a target historical intention to send to the user, such as a commodity recommendation demand, a commodity detail consultation demand.

[0122] 1) The user adopts the target historical intention, and uses the decision scheme corresponding to the target historical intention to perform the corresponding operation;

[0123] 2) The user does not adopt the target historical intention or is not satisfied with the intention result, and needs to send similar intention inquiry information to the user, such as “What business do you want to consult?”, and then receives the real intention of the user this time. In some cases, the real intention and the system intention name may be different, or some intentions may cover other intentions in addition to the real intention of the user, so another historical intention that is most similar or has the highest correlation with the real intention of the user needs to be selected from multiple historical intentions, and then the decision scheme corresponding to the historical intention is used to perform the corresponding operation.

[0124] 3) The above target historical intention / similar historical intention and the corresponding decision scheme are injected into the highlighted node.

[0125] The method provided by the above embodiment is used in the e-commerce field, predicts the consultation intention of a user, and automatically pushes out a decision scheme, and is a process that must be implemented in a mixed business strong scene such as recommendation, consultation, after-sales, and logistics. The accuracy of identifying the consultation purpose of a user and the solution rate of product recommendation are improved through the construction of a behavior trajectory graph between users to assist in intention auditing and solving.

[0126] Referring to Figure 6 , a main module schematic diagram of a consultation processing device 600 provided by an embodiment of the application is shown, which comprises:

[0127] A positioning module 601 is configured to receive a user consultation behavior, capture behavior trajectory information and access frequency information of a user accessing a page, and locate a node currently located by the user in a behavior trajectory graph; wherein the behavior trajectory graph is formed by connecting a plurality of nodes.

[0128] A disassembling module 602 is configured to disassemble the structure of the behavior trajectory graph to determine the level of the node and retrieve historical intentions corresponding to other nodes in each level except the node.

[0129] A screening module 603 is configured to calculate the effective value of each historical intention, screen a target historical intention according to the level and the effective value, and perform corresponding operations using a decision scheme corresponding to the target historical intention.

[0130] In the device, the positioning module 601 is configured to:

[0131] Starting from the time when the user initiates the consultation, chat information is captured for rough positioning to find basic information of the user; wherein the chat information comprises consultation content and reply content.

[0132] Based on the basic information of the user, behavior trajectory information and access frequency information of the user accessing a page are captured; wherein the behavior trajectory information is generated by capturing the trajectory of the user browsing different pages, and the access frequency information is generated by counting the number of times of the user accessing each page.

[0133] In the device, the positioning module 601 is configured to:

[0134] All historical behavior trajectories corresponding to the username of the user within a preset time period from the current time are queried, and then it is queried whether the all historical behavior trajectories exist in the behavior trajectory graph.

[0135] If the all historical behavior trajectories exist, a node currently located by the user in the behavior trajectory graph is located, otherwise the all historical behavior trajectories are added to the behavior trajectory graph to obtain a new behavior trajectory graph, and a new node is inserted; and

[0136] Detecting whether the access frequency information is contained under the node / new node, and adding the access frequency information under the node / new node if the access frequency information is not contained.

[0137] The device further comprises highlighting the node / new node in the behavior trajectory graph.

[0138] In the device, the disassembling module 602 is configured to:

[0139] Determine connection nodes connected to the node in the behavior trajectory graph, and for any node in the connection nodes, according to the number of connection lines of the any node, the any node is classified into a corresponding number of Ks layers.

[0140] For the node, after removing the connection nodes classified into the Ks=1 layer, it is judged whether the remaining connection nodes are in the same Ks layer, if yes, according to the number of the remaining connection nodes, the node is classified into a corresponding number of Ks layers; or

[0141] If not, the connection nodes classified into the Ks=2 layer are continuously removed, it is judged whether the remaining connection nodes are in the same Ks layer, if yes, according to the number of the remaining connection nodes, the node is classified into a corresponding number of Ks layers.

[0142] The above process is repeated to classify each connection node and each node into a corresponding Ks layer.

[0143] In the device, the disassembling module 602 is further configured to:

[0144] For any node in the connection nodes, the number of associations between the any node and other remaining nodes in the behavior trajectory graph is counted.

[0145] The ratio of the association number to the number of the other remaining nodes is taken as the degree of the any node, if the degree is 1, the any node is removed, otherwise, the any node is reserved.

[0146] The degree calculation and processing process is repeated until there is no connection node with a degree of 1.

[0147] In the device, the disassembling module 602 is further configured to: merge the nodes with the same name in the behavior trajectory graph.

[0148] In the device, the disassembling module 602 is configured to: retrieve historical users corresponding to other nodes, and based on the historical behavior trajectory of each historical user, obtain a corresponding historical intention.

[0149] The screening module 603 is configured to: calculate the effective value of each historical intention by using a graph classification method, and filter the historical intention with an effective value lower than an effective value threshold.

[0150] The screening module 603 is further configured to:

[0151] send the target historical intention to the user, and if an adoption message input by the user is received, perform corresponding operation using a decision scheme corresponding to the target historical intention; or

[0152] if no adoption message is received, send intention inquiry information to the user, then receive a real intention input by the user, find a historical intention with the highest similarity / association degree to the real intention from the historical intentions, and then perform corresponding operation using a decision scheme corresponding to the historical intention.

[0153] The apparatus further comprises: injecting the target historical intention and the decision scheme into the node.

[0154] In addition, the specific implementation of the apparatus in the embodiments of the present application has been described in detail in the above method, and thus the repeated content will not be described here.

[0155] Figure 7 An exemplary system architecture 700 to which embodiments of the application can be applied is shown, including terminal devices 701, 702, 703, a network 704, and a server 705 (only examples).

[0156] The terminal devices 701, 702, 703 can be various electronic devices with a display screen and supporting web browsing, installed with various communication client applications, and the user can use the terminal devices 701, 702, 703 to interact with the server 705 through the network 704 to receive or send messages, etc.

[0157] The network 704 is a medium for providing a communication link between the terminal devices 701, 702, 703 and the server 705. The network 704 can include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0158] The server 705 can be a server providing various services, configured to perform receiving user consultation behavior, capturing user behavior trajectory information and access frequency information, positioning the node where it is located, querying historical intentions and calculating effective values for screening operation.

[0159] It should be noted that the method provided by the embodiments of the present application is generally executed by the server 705, and accordingly, the apparatus is generally provided in the server 705.

[0160] It should be understood that, Figure 7 The number of terminal devices, networks and servers in the system architecture 700 is only illustrative. According to the needs of implementation, there can be any number of terminal devices, networks and servers.

[0161] Reference will now be made to the drawings Figure 8 which show a schematic configuration of a computer system 800 suitable for use in implementing the terminal device of an embodiment of the present application. Figure 8 The terminal device shown is merely an example and should not impose any limitation on the functions and the range of use of an embodiment of the present application.

[0162] As shown in Figure 7 , the computer system 800 includes a central processing unit (CPU) 801 which can perform various appropriate actions and processes in accordance with a program stored in a read only memory (ROM) 802 or a program loaded from a storage section 808 into a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the system 800 are also stored. The CPU 801, the ROM 802, and the RAM 803 are connected to each other through a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0163] The following components are connected to the I / O interface 805: an input section 806 including a keyboard, a mouse, etc.; an output section 807 including a display such as a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 808 including a hard disk, etc.; and a communication section 809 including a network interface card such as a LAN card, a modem, etc. The communication section 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to the I / O interface 805 as necessary. A removable recording medium 811 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc. is attached to the drive 810 as necessary, so that a computer program read therefrom is installed into the storage section 808 as necessary.

[0164] In particular, the processes described above with reference to the flowcharts can be implemented as a computer software program in accordance with an embodiment of the present application. For example, an embodiment of the present application includes a computer program product comprising a computer program carried on a computer readable medium, the computer program containing program code for executing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from a network by the communication section 809 and / or installed from the removable recording medium 811. When the computer program is executed by the central processing unit (CPU) 801, the above-described functions defined in the system of the present application are performed.

[0165] It should be noted that the computer-readable medium shown in the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination of the above. More specific examples of computer-readable storage media can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, 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 disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or component. In the present application, the computer-readable signal medium can include a data signal carried in a baseband or as a part of a carrier wave, which carries computer-readable program code. Such a propagated data signal can take various forms, including but not limited to an electromagnetic signal, an optical signal or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium, which can send, propagate or transmit a program for use by or in conjunction with an instruction execution system, device or component. The program code contained on the computer-readable medium can be transmitted by any suitable medium, including but not limited to wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0166] The flowcharts and block diagrams in the drawings illustrate the possible implementation architectures, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowcharts or block diagrams can represent a module, a program segment or a part of code containing one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur in different order than that shown in the drawings. For example, two blocks that are shown in succession can actually be executed substantially in parallel, and sometimes in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams or flowcharts, and the combination of blocks in the block diagrams or flowcharts, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0167] The modules described in the embodiments of the present application can be implemented in the form of software or hardware. The described modules can also be arranged in a processor, for example, a processor can be described as including a positioning module, a disassembling module, and a screening module. In some cases, the names of these modules do not constitute a limitation on the modules themselves, for example, the screening module can also be described as a "historical intention screening module".

[0168] As another aspect, the present application also provides a computer readable medium, which can be included in the device described in the above embodiments, or can exist independently without being assembled into the device. The computer readable medium carries one or more programs, which, when executed by the device, cause the device to include:

[0169] Receiving user consultation behavior, capturing user access page behavior track information and access frequency information to locate the node where the user is currently located in the behavior track graph; wherein the behavior track graph is composed of a plurality of nodes connected by lines;

[0170] Disassembling the structure of the behavior track graph to determine the level of the node, and retrieving the historical intention corresponding to other nodes in each level except the node;

[0171] Calculating the effective value of each historical intention, screening the target historical intention according to the level and the effective value size, and using the decision scheme corresponding to the target historical intention to perform corresponding operation.

[0172] According to the technical scheme of the embodiments of the present application, 1) for the dispersion problem of the system, a user-associated historical access complementary architecture model is provided, which can automatically learn the possible similar demands between users and associate the users, and use big data mechanism for comparative learning, and the similar intentions between users can also be directly recommended for solution. 2) for the redundancy problem of the system, a new intention fission and automatic scheme generation solution is provided, which can effectively calculate the user's intention and the existing intention through intention calculation in the intention recognition process. If it is found that the existing intention cannot cover all possible intentions, it will be summarized as a new intention method. Then, through user behavior analysis and related results in the user group, possible solutions are given, and they are split into related possible solutions.

[0173] The foregoing detailed description has set forth various embodiments of the devices and / or processes via the use of specific terminology. However, embodiments thereof can be practiced with the exact description not being presented in detail. The term "device" should be understood to encompass devices operating in various modes, such as active mode, sleep mode, hibernate mode, and the like. The terms "coupled" and "connected," along with derivatives thereof, can be used. It should be understood that these terms are not intended as synonyms for each other. Rather, particular circuitry that can be said to be coupled to, or connected with, other circuitry can be some of the other circuitry that can communicate in some way, while other circuitry can be some of the other circuitry that can not be in communication with that particular circuitry.

Claims

1. A consultation processing method characterized by comprising: The method comprises the following steps: Receiving user consultation behavior, capturing user access page behavior track information and access frequency information to locate the node where the user currently stays in the behavior track graph; wherein the behavior track graph is composed of multiple nodes connected; the locating the node where the user currently stays in the behavior track graph comprises: if the behavior track graph contains the user's historical behavior track information, locating the node where the user currently stays in the behavior track graph; Decomposing the structure of the behavior track graph to determine the level of the node, and retrieving the historical intention corresponding to other nodes in each level except the node; Calculating the effective value of each historical intention, screening out the target historical intention according to the level and the effective value size, and using the decision scheme corresponding to the target historical intention to perform corresponding operation.

2. The method of claim 1, wherein, The capturing user access page behavior track information and access frequency information comprises: Starting from the time when the user initiates consultation, capturing chat information for rough positioning to find user basic information; wherein the chat information includes consultation content and reply content; Based on the user basic information, capturing user access page behavior track information and access frequency information; wherein the behavior track information is generated by capturing the user's browsing track of different pages, and the access frequency information is generated by counting the number of times the user accesses each page.

3. The method according to claim 1 or 2, characterized in that, The locating the node where the user currently stays in the behavior track graph comprises: Querying all historical behavior tracks corresponding to the user's username within a preset time period from the current time, and then querying whether the all historical behavior tracks exist in the behavior track graph; If the all historical behavior tracks exist, locating the node where the user currently stays in the behavior track graph, otherwise adding the all historical behavior tracks to the behavior track graph to obtain a new behavior track graph, and inserting a new node; and Detecting whether the access frequency information is contained under the node / new node, if not, adding the access frequency information under the node / new node.

4. The method of claim 3, wherein, Further comprising: Highlighting the node / new node in the behavior track graph.

5. The method of claim 1, wherein, The decomposing the structure of the behavior track graph to determine the level of the node comprises: Determining the connection nodes connected with the node in the behavior track graph, for any node in the connection nodes, according to the number of connection lines of the any node, the any node is classified into a corresponding number of Ks layers; wherein Ks is a node importance ranking index; For the node, after removing the connection nodes classified into Ks=1 layer, judging whether the remaining connection nodes are in the same Ks layer, if yes, according to the number of remaining connection nodes, the node is classified into a corresponding number of Ks layers; or If not, continue to remove the connection nodes classified into Ks=2 layer, judge whether the remaining connection nodes are in the same Ks layer, if yes, according to the number of remaining connection nodes, the node is classified into a corresponding number of Ks layers; Repeat the above process to classify each connection node and each node into a corresponding Ks layer.

6. The method of claim 5, wherein, After the determining the connection nodes connected with the node in the behavior track graph, further comprising: For any one of the connection nodes, count the number of associations between the any one node and the remaining nodes in the behavior trajectory graph except the any one node; Take the ratio of the number of associations and the number of the remaining nodes as the degree of the any one node, if the degree is 1, eliminate the any one node, otherwise, keep it; Repeat the degree calculation and processing process until there is no connection node with a degree of 1.

7. The method according to claim 5 or 6, characterized in that, Before the structure of the behavior trajectory graph is disassembled, the method further includes merging the nodes with the same name in the behavior trajectory graph.

8. The method according to claim 5 or 6, characterized in that, The searching of the historical intentions corresponding to the other nodes in each level includes: searching the historical users corresponding to the other nodes, and obtaining the corresponding historical intentions based on the historical behavior trajectories of each historical user; The calculation of the effective values of each historical intention includes: calculating the effective values of each historical intention in a graph classification manner, and filtering the historical intentions with effective values lower than an effective value threshold.

9. The method of claim 1, wherein, The using of the decision scheme corresponding to the target historical intention to perform the corresponding operation further includes: sending the target historical intention to the user, if an adoption message input by the user is received, using the decision scheme corresponding to the target historical intention to perform the corresponding operation; or if no adoption message is received, sending intention query information to the user, then receiving a real intention input by the user, finding a historical intention with the highest similarity / association degree to the real intention from the historical intentions, and then using the decision scheme corresponding to the one historical intention to perform the corresponding operation.

10. The method of claim 1, wherein, Further including: injecting the target historical intention and the decision scheme into the node.

11. A consultation processing apparatus characterized by comprising: Including: a positioning module, configured to receive user consultation behavior, capture behavior trajectory information and access frequency information of user access pages, and locate a node in which the user currently stays in a behavior trajectory graph; wherein the behavior trajectory graph is composed of a plurality of nodes connected; the locating of the node in which the user currently stays in the behavior trajectory graph includes: if the behavior trajectory graph contains historical behavior trajectory information of the user this time, locating the node in which the user currently stays in the behavior trajectory graph; a disassembly module, configured to disassemble the structure of the behavior trajectory graph to determine a level in which the node stays, and search historical intentions corresponding to other nodes in each level except the node; a screening module, configured to calculate effective values of each historical intention, screen a target historical intention according to the level and the effective value, and use a decision scheme corresponding to the target historical intention to perform a corresponding operation.

12. An electronic device, comprising: Including: one or more processors; a storage device configured to store one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the method in any one of claims 1-10.

13. A computer readable medium having stored thereon a computer program, characterized in that The program is executed by the processor to implement the method in any one of claims 1-10.

Images