CN114385816A - Conversation flow mining method and device, electronic equipment and computer storage medium - Google Patents
Conversation flow mining method and device, electronic equipment and computer storage medium Download PDFInfo
- Publication number
- CN114385816A CN114385816A CN202210032825.6A CN202210032825A CN114385816A CN 114385816 A CN114385816 A CN 114385816A CN 202210032825 A CN202210032825 A CN 202210032825A CN 114385816 A CN114385816 A CN 114385816A
- Authority
- CN
- China
- Prior art keywords
- node
- intention information
- virtual node
- initial
- cluster
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
- G06F16/358—Browsing; Visualisation therefor
-
- 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
- G06F16/3331—Query processing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the application provides a conversation flow mining method, a conversation flow mining device, electronic equipment and a computer storage medium, and provides a visual interface; responding to a trigger operation of a first virtual node in an interface, and determining a first cluster intention which has a mapping relation with a node intention of the first virtual node; carrying out intention clustering on the corresponding first dialogue file to obtain a second dialogue file and a second cluster intention; adding an initial node intention corresponding to the second cluster intention and a second initial virtual node corresponding to the initial node intention, and forming topological connection between virtual nodes by taking the second initial virtual node as a downstream node of the first virtual node; in response to the editing operation on the selected second initial virtual node, adjusting the second initial virtual node to a second updated virtual node; and establishing a mapping relation between the node intention of the second updating virtual node and the second cluster intention, and obtaining and displaying the node intention of the second updating virtual node based on the mapping relation. The efficiency of excavation is improved.
Description
Technical Field
The embodiment of the application relates to the technical field of internet, in particular to a conversation flow mining method and device, electronic equipment and a computer storage medium.
Background
At present, intelligent dialogue robots are widely applied to various scenes, and can dialogue with customers through natural language so as to complete specified tasks of the customers. In the actual operation process, the intelligent dialogue robot performs actual operation based on the dialogue flow mined in advance, specifically: the method comprises the steps of firstly obtaining a dialogue statement input by a client, identifying the intention of the statement, then positioning a node corresponding to the intention in a pre-constructed dialogue flow, further determining the next node (target node) of the node in the whole process, finally executing corresponding operation based on the intention corresponding to the target node, and outputting a corresponding response statement.
The conversation flow mining process is a process of clustering the intention of conversation data in multiple rounds. When a specific mining turn is triggered, intent clustering is performed on a group of dialogue data, so that a plurality of intent class clusters are obtained, and each intent class cluster corresponds to a group of sub-dialogue data with the same intent.
In order to improve mining efficiency, parallel mining is generally triggered independently for multiple groups of dialogue data, however, in each mining round, a user may perform manual editing operations such as merging and deleting on mining results, which may affect the mining results, and in order to ensure accuracy of the mining results, intent clustering needs to be performed again on the dialogue data on which the editing operations are performed, which may result in a rapid reduction in mining efficiency of the whole dialogue flow.
Disclosure of Invention
In view of the above, embodiments of the present application provide a method for mining dialog flow to at least partially solve the above problem.
According to a first aspect of the embodiments of the present application, there is provided a method for mining a dialog flow, which is used for mining an original dialog file, the method including:
providing a visual interface used for displaying at least one virtual node and node intention information of the at least one virtual node;
in response to a trigger operation on a first virtual node, determining first cluster intention information having a mapping relation with node intention information of the first virtual node;
performing intention clustering on the first dialogue files corresponding to the first category intention information to obtain a plurality of second dialogue files and second category intention information corresponding to each second dialogue file;
adding initial node intention information corresponding to the second cluster intention information respectively and second initial virtual nodes corresponding to the initial node intention information respectively in a visual interface, and taking the second initial virtual nodes as downstream nodes of the first virtual nodes to form a topological connection relation among the virtual nodes;
receiving user editing operation input for one or more selected second initial virtual nodes;
responding to the editing operation of the selected second initial virtual nodes, and adjusting each second initial virtual node into a second updated virtual node so as to update the topological connection relation;
and establishing a mapping relation between the node intention information of each second updating virtual node and the second cluster intention information, and obtaining and displaying the node intention information of each second updating virtual node based on the mapping relation.
According to a second aspect of the embodiments of the present application, there is provided a conversation flow mining apparatus for mining an original conversation file, including:
the visual interface providing module is used for providing a visual interface which is used for displaying at least one virtual node and node intention information of the at least one virtual node;
the first cluster intention information determining module is used for responding to triggering operation of a first virtual node and determining first cluster intention information which has a mapping relation with the node intention information of the first virtual node;
the intention clustering module is used for carrying out intention clustering on the first dialogue files corresponding to the first dialogue file type intention information to obtain a plurality of second dialogue files and second dialogue file type intention information corresponding to each second dialogue file;
a topology connection relation obtaining module, configured to add, in a visual interface, initial node intention information corresponding to each second-class cluster intention information, and second initial virtual nodes corresponding to each initial node intention information, and use the second initial virtual nodes as downstream nodes of the first virtual nodes to form a topology connection relation between the virtual nodes;
the receiving module is used for receiving the editing operation input of the user on the selected one or more second initial virtual nodes;
the topological connection relation updating module is used for responding to the editing operation of the selected second initial virtual nodes and adjusting each second initial virtual node into a second updated virtual node so as to update the topological connection relation;
and the mapping relation establishing and node intention information displaying module is used for establishing a mapping relation between the node intention information of each second updating virtual node and the second cluster intention information, and obtaining and displaying the node intention information of each second updating virtual node based on the mapping relation.
According to a third aspect of embodiments of the present application, there is provided an electronic apparatus, including: the system comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete mutual communication through the communication bus; the memory is used for storing at least one executable instruction, and the executable instruction causes the processor to execute the operation corresponding to the conversation flow mining method according to the first aspect.
According to a fourth aspect of embodiments of the present application, there is provided a computer storage medium storing a computer program for live interaction, the computer program being stored on a computer storage medium and when executed by a processor, implementing a method for conversational flow mining of a live broadcast room as described in the first aspect.
According to a fifth aspect of the embodiments of the present application, there is provided a computer program product for live broadcast interaction, including computer instructions, where the computer instructions instruct a computing device to execute an operation corresponding to the live broadcast.
According to the conversation flow mining method provided by the embodiment of the application, data related to a conversation flow mining process is divided into 4 levels: the system comprises a virtual node level, a node intention information level, a class cluster intention information level and a dialogue file level, wherein the levels have the following association relationship in sequence: the virtual nodes in the virtual node hierarchy correspond to the node intention information in the node intention information hierarchy one by one; the node intention information in the node intention information hierarchy has a mapping relation with the class cluster intention information in the class cluster intention information hierarchy; the cluster-like intention information in the cluster-like intention information hierarchy corresponds to the dialog files in the dialog file hierarchy one to one.
Displaying virtual nodes in a virtual node hierarchy and corresponding node intention information in a node intention information hierarchy in a visual interface so as to allow a user to carry out an interactive mining process; when the first virtual node is triggered, namely mining starts, according to the association relationship, sequentially determining: the first cluster intention information and the corresponding first dialogue file have a mapping relation with the node intention information of the node; performing intention clustering on the first dialogue file to obtain a second dialogue file and corresponding second cluster intention information, and adding initial node intention information corresponding to the second cluster intention information and a corresponding second initial virtual node (a downstream node or a child node of the first virtual node) in an interface to form a preliminary mining result; when the selected second initial virtual node is triggered to be edited, namely when the user corrects the initial mining result, only the mapping relation among the virtual nodes in the virtual node hierarchy, the node intention information in the corresponding node intention information hierarchy, the corresponding node intention information and the cluster intention information in the cluster intention information hierarchy is adjusted, and the dialog files in the dialog file hierarchy do not need to be subjected to intention clustering again, so that the efficiency of dialog flow mining is improved.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments described in the embodiments of the present application, and other drawings can be obtained by those skilled in the art according to the drawings.
Fig. 1 is a schematic diagram of a management structure of visual conversation flow data mining according to a first embodiment of the present application;
fig. 2 is a schematic view of a scene of a conversation flow mining method according to a first embodiment of the present application;
fig. 3 is a schematic flowchart of mining task triggering of visual conversation flow mining according to a first embodiment of the present application;
FIG. 4 is a schematic diagram illustrating an editing operation performed on a node during an interactive annotation process according to a first embodiment of the present application;
fig. 5 is a schematic diagram of a multi-round mining process of visual dialog flow mining according to a first embodiment of the present application;
FIG. 6 is a flowchart illustrating steps of a method for mining dialog flow according to a first embodiment of the present application;
fig. 7 is a block diagram illustrating a structure of a conversation flow mining apparatus according to a second embodiment of the present application;
fig. 8 is a schematic structural diagram of an electronic device according to a third embodiment of the present application.
Detailed Description
In order to make those skilled in the art better understand the technical solutions in the embodiments of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, but not all embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present application shall fall within the scope of the protection of the embodiments in the present application.
The following further describes specific implementations of embodiments of the present application with reference to the drawings of the embodiments of the present application.
The first embodiment,
Referring to fig. 1, fig. 1 is a schematic view of a management structure of visual conversation flow data mining provided in an embodiment of the present application, and to facilitate understanding of a technical solution of the present application, first, a description is given, with reference to fig. 1, of the management structure of visual conversation flow data mining in the embodiment of the present application.
In the embodiment of the application, the mining abstraction of the conversation flow data is four layers: the system comprises a virtual node layer, a node intention information layer, a cluster-like intention information layer and a dialogue file layer. The virtual node layer comprises one or more virtual nodes, and when a plurality of virtual nodes exist, the virtual node layer also comprises a topological connection relation among the virtual nodes, and the topological connection relation is used for showing to a user in a visual interface. The node intention information layer comprises node intention information which is in one-to-one correspondence with each virtual node in the virtual node layer, the node intention information is used for representing the intention of the corresponding virtual node, and the hierarchy is also displayed in the visual interface. The class cluster intention information layer comprises class cluster intention information, and one-to-one or one-to-many mapping relation exists between the node intention information and the class cluster intention information. Specifically, one piece of node intention information may have a mapping relationship with only one piece of class cluster intention information (in this case, the node intention information may be the same as the class cluster intention information), or one piece of node intention information may have a mapping relationship with a plurality of pieces of class cluster intention information at the same time (in this case, the node intention information may be intention information obtained after aggregating the various pieces of class cluster intention information). The dialog file layer comprises dialog files corresponding to the class cluster intention information one by one, namely, after the intention is clustered, the dialog files with the same intention are used as a class cluster, the specific dialog files are stored in the dialog file layer for being used by the dialog mining task, and the intention of the class cluster is stored in the class cluster intention information layer as the class cluster intention information.
The following provides a general description of the conversation flow mining process of the embodiment of the present application on the basis of the management structure of fig. 1. Assume that after a first round of mining on original virtual node a0, two first virtual nodes are formed in the virtual node layer: a11 and a12, for the first virtual node a11, there is a node intention information b11 corresponding to the node intention information layer, and the node intention information b11 has a mapping relation with the first cluster intention information c11, c12 and c13 in the cluster intention information layer. The first cluster intention information c11, c12, and c13 correspond to d11, d12, and d13 in the dialog file layer, respectively. When the first virtual node a11 is triggered to perform the second round of mining, d11, d12 and d13 in the dialog file layer are taken as a whole to be clustered, and intent clustering is performed, so that a corresponding second dialog file d2 is obtained. Correspondingly, the second type cluster intention information c2 corresponding to the second dialogue file d2 exists in the class cluster intention information layer, the node intention information b2 having a mapping relation with the second type cluster intention information c2 exists in the node intention information layer, and the second virtual node a2 corresponding to the node intention information b2 exists in the virtual node layer. And then, according to the needs, analogizing in sequence, triggering the mining subprocesses of the subsequent rounds such as the 3 rd round and the like for the virtual nodes in the virtual node layer until the mining process is finished, and the details are not repeated here.
Referring to fig. 2, fig. 2 is a schematic view of a scenario of a dialog flow mining method according to a first embodiment of the present application, and for convenience of understanding, an application scenario of the dialog flow mining method according to the first embodiment of the present application is first explained with reference to fig. 2.
The embodiment of the present application provides a visual interface as shown in fig. 2 (a), where the visual interface is used to show at least one virtual node (a first virtual node is taken as an example in the figure and does not constitute a limitation on the number of virtual nodes in the embodiment of the present application) and node intention information of the at least one virtual node, where the first virtual node is in a virtual node layer in fig. 1, and corresponding node intention information is in a node intention information layer in fig. 1. When a user performs a trigger operation on a first virtual node, determining first cluster intention information (in a cluster intention information layer in fig. 2) having a mapping relation with node intention information of the first virtual node, and a first dialog file (in a dialog file layer in fig. 2) corresponding to the first cluster intention information. Then, the first dialog files are subjected to intent clustering to obtain 2 second dialog files (only 2 second dialog files are taken as an example in the figure, and no limitation is made to the number of the second dialog files), and second-class cluster intent information corresponding to each second dialog file. Then, as shown in fig. 2 (b), initial node intention information corresponding to each second-class cluster intention information and second initial virtual nodes corresponding to each initial node intention information are added to the visual interface, and each second initial virtual node is used as a downstream node of the first virtual node to form a topological connection relationship between the virtual nodes. When receiving an input of an editing operation (e.g., a merging operation) by a user for 2 second initial virtual nodes (which are illustrated by taking 2 second initial virtual nodes as an example in the figure and do not constitute a limitation on the number of selected second initial virtual nodes) in the visualization interface shown in (b) in fig. 2, as shown in (c) in fig. 2, adjusting the second initial virtual nodes to be second updated virtual nodes so as to update the topological connection relationship; and establishing a mapping relation between the node intention information of the second updating virtual nodes and the second cluster intention information, and obtaining and displaying the node intention information of each second updating virtual node based on the mapping relation.
It should be noted that, in the embodiment of the present application, in the visualization interface, a specific display form of the topological connection relationship between the virtual nodes is not limited, for example, the topological connection relationship may be displayed in a flow chart form as shown in fig. 2, or may be displayed in other forms such as a table, and the flow chart form is merely illustrated in fig. 2, and does not limit the embodiment of the present application.
The following presents a visualization dialog flow mining process according to an embodiment of the present application. Referring to fig. 3 to 5, a visual interactive interface may be provided to show one or more virtual nodes and corresponding node intention information for a user to perform mining or editing operations. In the visual dialog flow mining process, multiple rounds of interactive mining sub-processes are generally required to be performed on the original dialog file to generate the final dialog flow of the task-based dialog. With respect to a single interactive mining sub-process, two steps are typically involved: the first step, triggering of a mining task; and step two, interactive labeling. Referring to fig. 3, the first step is: when a user performs a trigger operation on a certain node (such as a start node) through a visual interface, the dialog files corresponding to the node are automatically subjected to intention clustering, so that a plurality of dialog subfiles and corresponding child nodes (such as nodes 1 to 5) are obtained. In the second step, the user may perform manual intervention labeling on the mining result through the visual interactive interface, that is, perform an editing operation on the node, so as to obtain a labeled mining result, where, referring to fig. 4, the editing operation on the node may include: merge operations, delete operations, etc.
Referring to fig. 5, fig. 5 is a schematic diagram of a multi-turn dialog flow mining process, in which 2-turn interactive mining sub-processes are collectively performed. In the 1 st turn, the user may perform a merge operation on the visual interactive interface, so that the newly generated node 2 and node 3 are merged into a node 3 ', and simultaneously perform a delete operation to delete the node 5, thereby obtaining a node 1, a node 3', and a node 4. In the 2 nd round, the user may trigger the mining operations of node 1, node 3 ', and node 4, respectively, to generate node 1A and node 1B based on node 1, node 3 ' a based on node 3 ', and node 4A based on node 4.
Referring to fig. 6, fig. 6 is a flowchart illustrating steps of a conversation flow mining method according to a first embodiment of the present disclosure; specifically, the dialog flow mining method provided in this embodiment is used to mine an original dialog file, where the original dialog file may refer to multiple dialog sentences input by a client, and each dialog sentence may include multiple turns of dialog data.
The conversation flow mining method comprises the following steps:
step 602, providing a visual interface, where the visual interface is used to show at least one virtual node and node intention information of the at least one virtual node.
In the embodiment of the application, a one-to-one correspondence relationship exists between the virtual nodes in the visual interface and the node intention information of the virtual nodes. The node intent information of a virtual node characterizes the intent that the virtual node represents.
For example: the original dialog file is a dialog file between the customer service and the client generated in the air ticket purchasing scene, and the node intention information can be travel time inquiry, travel number inquiry, traveler identity information inquiry, air ticket type inquiry and the like.
In the embodiment of the present application, how to perform the trigger operation on the first virtual node is not limited, for example: the operations such as clicking and dragging of the first virtual node displayed in the visual interface can be used as trigger operations, and therefore the mining process of the conversation flow is triggered.
As indicated in the above description of fig. 1, in the embodiment of the present application, there may be a one-to-one or one-to-many mapping relationship between the node intention information of the virtual node and the class cluster intention information. Specifically, the node intention information of one virtual node may have a mapping relation with only one class cluster intention information, or may have a mapping relation with a plurality of class cluster intention information at the same time. When the node intention information of a virtual node has a mapping relationship with multiple cluster intention information at the same time, the node intention information of the virtual node may be aggregated from the multiple cluster intention information, for example, the node intention information of the virtual node is: the class cluster intention information with which the air ticket is purchased and which has a mapping relation may include 2, which are respectively: purchase tickets for airline a and ticket for airline B.
The second type of cluster intention information corresponds to the second dialog file one to one, namely: and a second dialogue file corresponds to a second cluster intention information, and the second cluster intention information represents the intention corresponding to the second dialogue file.
Further, in this embodiment of the application, after obtaining a plurality of second session files, each second session file may be stored separately.
In this step, for each piece of second-class cluster intention information obtained in step 606, corresponding initial node intention information and a corresponding second initial virtual node may be added to the visual interface, and the newly added second initial virtual node is used as a child node (downstream node) of the original first virtual node in the visual interface, so as to form a topological connection relationship between the first virtual node and the second initial virtual node.
The initial node intention information displayed in the visual interface, the second initial virtual node displayed in the visual interface and the second cluster intention information which is not displayed in the visual interface form a one-to-one correspondence relationship.
Steps 610-614 are explained below:
after step 608, the second initial virtual nodes formed in the visual interface are all virtual nodes corresponding to the second type of cluster intention information obtained by the automatic intention clustering operation, that is, the mining process is automatically completed without manual correction. However, according to different service categories actually provided by the user, the user may be required to perform annotation correction on the mining result, for example, according to the initial node intention information, edit operations such as merging, deleting, and restoring are performed on the second initial virtual node to obtain second updated virtual nodes, and a mapping relationship between the node intention information of each second updated virtual node and the second-class cluster intention information is adjusted.
Taking an air ticket purchasing scenario as an example, suppose that 2 second initial virtual nodes are newly added in the visual interface through step 608, and the corresponding initial node intention information is: and purchasing tickets of the A airline company and purchasing tickets of the B airline company, wherein the tickets of the A airline company and the tickets of the B airline company are not substantially different in service processing flow by combining the service conditions actually provided by the customer, and at the moment, the user can perform merging operation on 2 second initial virtual nodes, so that the second initial virtual nodes are updated into second updated virtual nodes in a visual interface, and the mapping relation between the node intention information of the second updated virtual nodes and the second cluster intention information is adjusted.
For another example, it is assumed that the initial node intention information corresponding to the 2 second initial virtual nodes is: the ticket is purchased and the check-in service is transacted, but the service actually provided by the client does not include the check-in service at present. At this time, the user may perform a deletion operation on the second initial virtual node corresponding to the check-in service. Subsequently, with the continuous expansion of the service types provided by the customers, the value machine service can be increased. At this time, the user may perform a recovery operation on the deleted second initial virtual node corresponding to the check-in service, and adjust a mapping relationship between the node intention information of the second initial virtual node after recovery and the second-class cluster intention information.
Further, in this embodiment of the application, the editing operation is only used to adjust a mapping relationship between the node intention information of each second update virtual node and the second-class cluster intention information, and does not change the storage states of the plurality of second session files. For example, for the merge operation, only the mapping relationship between the node intention information of the second updated virtual node and the second type of cluster intention information is adjusted, and the second dialog file itself is not merged. For the deletion operation, only the second initial virtual node to be deleted and the corresponding initial node intention information are deleted, and any second session file itself is not deleted.
Optionally, in some embodiments, the mapping relationship between the node intention information of each second updated virtual node and the second type of cluster intention information may be adjusted based on the specific type of the editing operation.
Specifically, for the merge operation, the obtaining process of the second updated virtual node and the adjusting process of the mapping relationship between the node intention information of the second updated virtual node and the second class cluster intention information may include:
in response to the combination operation of the selected plurality of second initial virtual nodes, deleting the selected plurality of second initial virtual nodes, and adding the combined virtual nodes in the visual interface; the unselected second initial virtual nodes and the merged virtual nodes are used as second updated virtual nodes;
if the second updated virtual node is the merged virtual node, determining second cluster intention information corresponding to each selected second initial virtual node as second cluster intention information having a mapping relation with the second updated virtual node; aggregating second cluster intention information corresponding to each selected second initial virtual node to obtain node intention information of the second updated virtual node and displaying the node intention information;
if the second updated virtual node is a non-selected second initial virtual node, the initial node intention information corresponding to the non-selected second initial virtual node is used as the node intention information of the second updated virtual node; and determining the second type of cluster intention information corresponding to the unselected second initial virtual node as the second type of cluster intention information having a mapping relation with the node intention information of the second updated virtual node.
For the deletion operation, the obtaining process of the second updated virtual node and the adjusting process of the mapping relationship between the node intention information of the second updated virtual node and the second type cluster intention information may include:
in response to the deletion operation of the selected second initial virtual node, deleting the selected second initial virtual node and the initial node intention information corresponding to the selected second initial virtual node in the visual interface; reserving second cluster intention information corresponding to the selected second initial virtual node as soft deletion second cluster intention information; reserving a second dialogue file corresponding to the second cluster intention information; determining the non-selected second initial virtual node as a second updated virtual node; taking the initial node intention information corresponding to each unselected second initial virtual node as the node intention information of each second updated virtual node; and respectively determining second cluster intention information corresponding to the non-selected second initial virtual nodes as second cluster intention information having a mapping relation with the node intention information of the second updated virtual nodes.
In addition, for the recovery operation, the virtual node updating process and the adjusting process of the mapping relationship between the node intention information of the updated virtual node and the second type of cluster intention information may include:
responding to the recovery operation of the deleted second initial virtual node, and determining soft deletion second cluster intention information corresponding to the deleted second initial virtual node; adding the recovered second virtual node in the visual interface; determining soft deletion second-class cluster intention information corresponding to the deleted second initial virtual node as a second-class cluster intention which has a mapping relation with the node intention information of the recovered second virtual node; and displaying the soft deletion second-type cluster intention information corresponding to the deleted second initial virtual node as the node intention information of the recovered second virtual node.
In the embodiment of the application, the virtual nodes in the virtual node hierarchy and the corresponding node intention information in the node intention information hierarchy can be displayed in the visual interface, so that a user can carry out an interactive mining process. When the first virtual node is triggered, namely mining starts, determining in sequence according to the association relationship: the first cluster intention information and the corresponding first dialogue file have a mapping relation with the node intention information of the node; and performing intention clustering on the first dialogue file to obtain a second dialogue file and corresponding second cluster intention information, and adding initial node intention information corresponding to the second cluster intention information and a corresponding second initial virtual node (a downstream node or a child node of the first virtual node) in an interface to form a preliminary mining result. When the selected second initial virtual node is triggered to be edited, namely when the user corrects the initial mining result, only the mapping relation among the virtual nodes in the virtual node hierarchy, the node intention information in the corresponding node intention information hierarchy, the corresponding node intention information and the cluster intention information in the cluster intention information hierarchy is adjusted, and the dialog files in the dialog file hierarchy do not need to be subjected to intention clustering again, so that the efficiency of dialog flow mining is improved.
Example II,
Referring to fig. 7, fig. 7 is a block diagram illustrating a structure of a conversation partner mining apparatus according to a second embodiment of the present application. The conversation flow mining device provided by the embodiment of the application comprises: a visual interface providing module 702, configured to provide a visual interface, where the visual interface is used to show at least one virtual node and node intention information of the at least one virtual node. A first cluster intention information determining module 704, configured to determine, in response to a trigger operation on a first virtual node, first cluster intention information having a mapping relationship with node intention information of the first virtual node. The intention clustering module 706 is configured to perform intention clustering on the first dialog files corresponding to the first category of intention information to obtain a plurality of second dialog files and second category of intention information corresponding to each second dialog file. A topology connection relation obtaining module 708, configured to add, in the visualization interface, the initial node intention information corresponding to each second-class cluster intention information and the second initial virtual nodes corresponding to each initial node intention information, and use the second initial virtual nodes as downstream nodes of the first virtual nodes to form a topology connection relation between the virtual nodes. A receiving module 710, configured to receive an editing operation input of the user on the selected one or more second initial virtual nodes. And a topology connection relation updating module 712, configured to adjust each second initial virtual node to a second updated virtual node in response to the editing operation on the selected second initial virtual node, so as to update the topology connection relation. And a mapping relationship establishing and node intention information displaying module 714, configured to establish a mapping relationship between the node intention information of each second updated virtual node and the second-class cluster intention information, and obtain and display the node intention information of each second updated virtual node based on the mapping relationship.
Optionally, in some embodiments, when the step of establishing the mapping relationship between the node intention information of each second updated virtual node and the second cluster intention information is executed, the mapping relationship establishing and node intention information displaying module 714 is specifically configured to: and adjusting the mapping relation between the node intention information of each second updating virtual node and the second cluster intention information based on the type of the editing operation.
Optionally, in some embodiments, the conversation flow mining apparatus further includes: and the storage module is used for respectively storing the plurality of second dialogue files after the plurality of second dialogue files are obtained.
Optionally, in some embodiments, the editing operation is used to adjust a mapping relationship between the node intention information of each second updated virtual node and the second type of cluster intention information, without changing the storage states of the plurality of second session files.
Optionally, in some embodiments, the type of editing operation includes: merge operations, delete operations, or restore operations.
Optionally, in some embodiments, the topology connection relationship updating module 712 is specifically configured to: in response to the combination operation of the selected plurality of second initial virtual nodes, deleting the selected plurality of second initial virtual nodes, and adding the combined virtual nodes in the visual interface; and using the unselected second initial virtual node and the merged virtual node as a second updated virtual node.
Optionally, in some embodiments, the mapping relationship establishing and node intention information displaying module 714 is specifically configured to: if the second updated virtual node is the merged virtual node, determining second cluster intention information corresponding to each selected second initial virtual node as second cluster intention information having a mapping relation with the second updated virtual node; and aggregating the second cluster intention information corresponding to each selected second initial virtual node to obtain the node intention information of the second updated virtual node and displaying the node intention information.
Optionally, in some embodiments, the topology connection relationship updating module 712 is specifically configured to: in response to the deletion operation of the selected second initial virtual node, deleting the selected second initial virtual node and the initial node intention information corresponding to the selected second initial virtual node in the visual interface; reserving second cluster intention information corresponding to the selected second initial virtual node as soft deletion second cluster intention information; reserving a second dialogue file corresponding to the second cluster intention information; determining the non-selected second initial virtual node as a second updated virtual node; a mapping relationship establishing and node intention information displaying module 714, configured to specifically use initial node intention information corresponding to each unselected second initial virtual node as node intention information of each second updated virtual node; and respectively determining second cluster intention information corresponding to the non-selected second initial virtual nodes as second cluster intention information having a mapping relation with the node intention information of the second updated virtual nodes.
Optionally, in some embodiments, the conversation flow mining apparatus further includes: the soft deletion second-type cluster intention information is used for responding to the recovery operation of the deleted second initial virtual node and determining the soft deletion second-type cluster intention information corresponding to the deleted second initial virtual node; the recovery node adding module is used for adding a recovered second virtual node in the visual interface; a mapping relation determining module, configured to determine soft deletion second-class cluster intention information corresponding to the deleted second initial virtual node as a second-class cluster intention having a mapping relation with the node intention information of the restored second virtual node; and the recovery node intention information display module is used for displaying the soft deletion second cluster intention information corresponding to the deleted second initial virtual node as the node intention information of the recovered second virtual node.
Optionally, in some embodiments, the topological connection relationship between the virtual nodes is shown in the visualization interface as: in flow chart form or in table form.
The dialog flow mining device of this embodiment is used to implement the corresponding dialog flow mining method in the foregoing method embodiments, and has the beneficial effects of the corresponding method embodiments, which are not described herein again. In addition, the functional implementation of each module in the dialog flow mining apparatus of this embodiment can refer to the description of the corresponding part in the foregoing method embodiment, and is not described herein again.
Example III,
Referring to fig. 8, a schematic structural diagram of an electronic device according to a third embodiment of the present application is shown, and the specific embodiment of the present application does not limit a specific implementation of the electronic device.
As shown in fig. 8, the electronic device may include: a processor (processor)802, a Communications Interface 804, a memory 806, and a communication bus 808. The processor 802, communication interface 804, and memory 806 communicate with one another via a communication bus 808. The communication interface 804 is used for communication with other electronic devices or servers. The processor 802 is configured to execute the program 810, and may specifically execute the relevant steps in the above-described dialog flow mining method embodiment. In particular, the program 810 may include program code comprising computer operating instructions.
The processor 802 may be a CPU, or an Application Specific Integrated Circuit (ASIC), or one or more Integrated circuits configured to implement embodiments of the present application. The intelligent device comprises one or more processors which can be the same type of processor, such as one or more CPUs; or may be different types of processors such as one or more CPUs and one or more ASICs.
The memory 806 is used to store a program 810. The memory 806 may comprise high-speed RAM memory, and may also include non-volatile memory (non-volatile memory), such as at least one disk memory.
The program 810 may be specifically configured to cause the processor 802 to perform the following operations: providing a visual interface, wherein the visual interface is used for displaying at least one virtual node and node intention information of the at least one virtual node; in response to a trigger operation on a first virtual node, determining first cluster intention information having a mapping relation with node intention information of the first virtual node; performing intention clustering on the first dialogue files corresponding to the first cluster intention information to obtain a plurality of second dialogue files and second cluster intention information corresponding to each second dialogue file; adding initial node intention information corresponding to the second cluster intention information respectively and second initial virtual nodes corresponding to the initial node intention information respectively in a visual interface, and taking the second initial virtual nodes as downstream nodes of the first virtual nodes to form a topological connection relation among the virtual nodes; receiving user editing operation input for one or more selected second initial virtual nodes; responding to the editing operation of the selected second initial virtual nodes, and adjusting each second initial virtual node into a second updated virtual node so as to update the topological connection relation; and establishing a mapping relation between the node intention information of each second updating virtual node and the second cluster intention information, and obtaining and displaying the node intention information of each second updating virtual node based on the mapping relation.
For specific implementation of each step in the program 810, reference may be made to corresponding steps and corresponding descriptions in units in the foregoing dialog flow mining method embodiment, which are not described herein again. It can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described devices and modules may refer to the corresponding process descriptions in the foregoing method embodiments, and are not described herein again.
Through the electronic equipment of the embodiment, virtual nodes in a virtual node hierarchy and corresponding node intention information in a node intention information hierarchy are displayed in a visual interface so that a user can carry out an interactive mining process; when the first virtual node is triggered, namely mining starts, according to the association relationship, sequentially determining: the first cluster intention information and the corresponding first dialogue file have a mapping relation with the node intention information of the node; performing intention clustering on the first dialogue file to obtain a second dialogue file and corresponding second cluster intention information, and adding initial node intention information corresponding to the second cluster intention information and a corresponding second initial virtual node (a downstream node or a child node of the first virtual node) in an interface to form a preliminary mining result; when the selected second initial virtual node is triggered to be edited, namely when the user corrects the initial mining result, only the mapping relation among the virtual nodes in the virtual node hierarchy, the node intention information in the corresponding node intention information hierarchy, the corresponding node intention information and the cluster intention information in the cluster intention information hierarchy is adjusted, and the dialog files in the dialog file hierarchy do not need to be subjected to intention clustering again, so that the efficiency of dialog flow mining is improved.
The present application further provides a computer storage medium storing a computer program for mining a conversation flow, on which the computer program is stored, and when the computer program is executed by a processor, the computer program implements any of the conversation flow mining methods in the above-described method embodiments.
The present application further provides a computer program product for mining a dialog flow, including computer instructions that instruct a computing device to perform an operation corresponding to any of the above-described method embodiments.
It should be noted that, according to the implementation requirement, each component/step described in the embodiment of the present application may be divided into more components/steps, and two or more components/steps or partial operations of the components/steps may also be combined into a new component/step to achieve the purpose of the embodiment of the present application.
The above-described methods according to embodiments of the present application may be implemented in hardware, firmware, or as software or computer code storable in a recording medium such as a CD ROM, a RAM, a floppy disk, a hard disk, or a magneto-optical disk, or as computer code originally stored in a remote recording medium or a non-transitory machine-readable medium downloaded through a network and to be stored in a local recording medium, so that the methods described herein may be stored in such software processes on a recording medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware such as an ASIC or FPGA. It will be appreciated that the computer, processor, microprocessor controller or programmable hardware includes memory components (e.g., RAM, ROM, flash memory, etc.) that can store or receive software or computer code that, when accessed and executed by the computer, processor or hardware, implements the conversation flow mining methods described herein. Further, when a general purpose computer accesses code for implementing the conversation flow mining methods illustrated herein, execution of the code transforms the general purpose computer into a special purpose computer for performing the conversation flow mining methods illustrated herein.
Those of ordinary skill in the art will appreciate that the various illustrative elements and method steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the embodiments of the present application.
The above embodiments are only used for illustrating the embodiments of the present application, and not for limiting the embodiments of the present application, and those skilled in the relevant art can make various changes and modifications without departing from the spirit and scope of the embodiments of the present application, so that all equivalent technical solutions also belong to the scope of the embodiments of the present application, and the scope of patent protection of the embodiments of the present application should be defined by the claims.
Claims (14)
1. A conversation flow mining method is used for mining an original conversation file, and comprises the following steps:
providing a visual interface used for displaying at least one virtual node and node intention information of the at least one virtual node;
in response to a trigger operation on a first virtual node, determining first cluster intention information having a mapping relation with node intention information of the first virtual node;
performing intention clustering on the first dialogue files corresponding to the first category intention information to obtain a plurality of second dialogue files and second category intention information corresponding to each second dialogue file;
adding initial node intention information corresponding to the second cluster intention information respectively and second initial virtual nodes corresponding to the initial node intention information respectively in a visual interface, and taking the second initial virtual nodes as downstream nodes of the first virtual nodes to form a topological connection relation among the virtual nodes;
receiving user editing operation input for one or more selected second initial virtual nodes;
responding to the editing operation of the selected second initial virtual nodes, and adjusting each second initial virtual node into a second updated virtual node so as to update the topological connection relation;
and establishing a mapping relation between the node intention information of each second updating virtual node and the second cluster intention information, and obtaining and displaying the node intention information of each second updating virtual node based on the mapping relation.
2. The method according to claim 1, wherein the establishing a mapping relationship between the node intention information of each second updated virtual node and the second cluster intention information comprises:
and adjusting the mapping relation between the node intention information and the second cluster intention information of each second updating virtual node based on the type of the editing operation.
3. The method of claim 2, wherein after the obtaining the plurality of second session files, the method further comprises:
and respectively storing the plurality of second dialogue files.
4. The method according to claim 3, wherein the editing operation is used to adjust a mapping relationship between node intention information and second type cluster intention information of each second updated virtual node without changing a storage state of the plurality of second session files.
5. The method of claim 3, wherein the type of editing operation comprises: merge operations, delete operations, or restore operations.
6. The method of claim 5, wherein said adjusting each second initial virtual node to a second updated virtual node in response to an edit operation to a selected second initial virtual node comprises:
in response to the combination operation of the selected plurality of second initial virtual nodes, deleting the selected plurality of second initial virtual nodes, and adding the combined virtual nodes in the visual interface;
and using the non-selected second initial virtual node and the merged virtual node as a second updated virtual node.
7. The method according to claim 6, wherein the adjusting the mapping relationship between the node intention information of each second updated virtual node and the second type cluster intention information based on the type of the editing operation, and obtaining and displaying the node intention information of each second updated virtual node based on the mapping relationship comprises:
if the second updated virtual node is the merged virtual node, determining second cluster intention information corresponding to each selected second initial virtual node as second cluster intention information having a mapping relation with the second updated virtual node;
and aggregating the second cluster intention information corresponding to each selected second initial virtual node to obtain the node intention information of the second updated virtual node and displaying the node intention information.
8. The method of claim 5, wherein said adjusting each second initial virtual node to a second updated virtual node in response to an edit operation to a selected second initial virtual node comprises:
in response to a deletion operation on a selected second initial virtual node, deleting the selected second initial virtual node and initial node intention information corresponding to the selected second initial virtual node in the visual interface;
reserving second cluster intention information corresponding to the selected second initial virtual node as soft deletion second cluster intention information; reserving a second dialogue file corresponding to the soft deletion second cluster intention information;
determining the non-selected second initial virtual node as a second updated virtual node;
the adjusting the mapping relationship between the node intention information of each second updated virtual node and the second cluster intention information based on the type of the editing operation, and obtaining and displaying the node intention information of each second updated virtual node based on the mapping relationship comprises:
taking the initial node intention information corresponding to each unselected second initial virtual node as the node intention information of each second updated virtual node;
and respectively determining second cluster intention information corresponding to the non-selected second initial virtual nodes as second cluster intention information having a mapping relation with the node intention information of the second updated virtual nodes.
9. The method of claim 8, wherein the method further comprises:
responding to the recovery operation of the deleted second initial virtual node, and determining soft deletion second cluster intention information corresponding to the deleted second initial virtual node;
adding the recovered second virtual node in the visual interface;
determining soft deletion second-class cluster intention information corresponding to the deleted second initial virtual node as a second-class cluster intention which has a mapping relation with the node intention information of the recovered second virtual node;
and displaying the soft deletion second cluster intention information corresponding to the deleted second initial virtual node as the node intention information of the recovered second virtual node.
10. The method according to claim 1, wherein the topological connection relationship among the virtual nodes is shown in the visual interface as: in flow chart form or in table form.
11. A conversation flow mining apparatus for mining an original conversation file, comprising:
the visual interface providing module is used for providing a visual interface which is used for displaying at least one virtual node and node intention information of the at least one virtual node;
the first cluster intention information determining module is used for responding to triggering operation of a first virtual node and determining first cluster intention information which has a mapping relation with the node intention information of the first virtual node;
the intention clustering module is used for carrying out intention clustering on the first dialogue files corresponding to the first dialogue file type intention information to obtain a plurality of second dialogue files and second dialogue file type intention information corresponding to each second dialogue file;
a topology connection relation obtaining module, configured to add, in a visual interface, initial node intention information corresponding to each second-class cluster intention information, and second initial virtual nodes corresponding to each initial node intention information, and use the second initial virtual nodes as downstream nodes of the first virtual nodes to form a topology connection relation between the virtual nodes;
the receiving module is used for receiving the editing operation input of the user on the selected one or more second initial virtual nodes;
the topological connection relation updating module is used for responding to the editing operation of the selected second initial virtual nodes and adjusting each second initial virtual node into a second updated virtual node so as to update the topological connection relation;
and the mapping relation establishing and node intention information displaying module is used for establishing a mapping relation between the node intention information of each second updating virtual node and the second cluster intention information, and obtaining and displaying the node intention information of each second updating virtual node based on the mapping relation.
12. An electronic device, comprising: the system comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete mutual communication through the communication bus;
the memory is used for storing at least one executable instruction, and the executable instruction causes the processor to execute the operation corresponding to the conversation flow mining method according to any one of claims 1 to 10.
13. A computer storage medium having stored thereon a computer program for dialog flow mining, the program, when being executed by a processor, implementing a dialog flow mining method according to any one of claims 1 to 10.
14. A computer program product for conversation flow mining, comprising computer instructions to instruct a computing device to perform operations corresponding to the conversation flow mining method of any one of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210032825.6A CN114385816A (en) | 2022-01-12 | 2022-01-12 | Conversation flow mining method and device, electronic equipment and computer storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210032825.6A CN114385816A (en) | 2022-01-12 | 2022-01-12 | Conversation flow mining method and device, electronic equipment and computer storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114385816A true CN114385816A (en) | 2022-04-22 |
Family
ID=81202579
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210032825.6A Pending CN114385816A (en) | 2022-01-12 | 2022-01-12 | Conversation flow mining method and device, electronic equipment and computer storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114385816A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115238060A (en) * | 2022-09-20 | 2022-10-25 | 支付宝(杭州)信息技术有限公司 | Man-machine interaction method and device, medium and computing equipment |
| CN116882408A (en) * | 2023-09-07 | 2023-10-13 | 南方电网数字电网研究院有限公司 | Construction method and device of transformer graph model, computer equipment and storage medium |
-
2022
- 2022-01-12 CN CN202210032825.6A patent/CN114385816A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115238060A (en) * | 2022-09-20 | 2022-10-25 | 支付宝(杭州)信息技术有限公司 | Man-machine interaction method and device, medium and computing equipment |
| CN116882408A (en) * | 2023-09-07 | 2023-10-13 | 南方电网数字电网研究院有限公司 | Construction method and device of transformer graph model, computer equipment and storage medium |
| CN116882408B (en) * | 2023-09-07 | 2024-02-27 | 南方电网数字电网研究院有限公司 | Construction method and device of transformer graph model, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108920135B (en) | User-defined service generation method and device, computer equipment and storage medium | |
| CN114385816A (en) | Conversation flow mining method and device, electronic equipment and computer storage medium | |
| CN109951547B (en) | Transaction request parallel processing method, device, equipment and medium | |
| CN112561386A (en) | Method, device and equipment for acquiring business process | |
| EP1868087A1 (en) | System and method for configuration application programs | |
| US8214245B2 (en) | Method and system for synchronizing inclusive decision branches | |
| CN115857920A (en) | Application page development method, device and system, computing equipment and storage medium | |
| CN109583746A (en) | The method and device of the routing rule of setting procedure, readable storage medium storing program for executing | |
| CN111722842A (en) | Pipeline arranging method and device, electronic equipment and storage medium | |
| CN115048254A (en) | Simulation test method, system, equipment and readable medium of data distribution strategy | |
| CN112148276A (en) | Visual programming for deep learning | |
| CN114157507A (en) | Cloud service vulnerability analysis method and artificial intelligence system adopting big data analysis | |
| US20120330859A1 (en) | Interactive business process modeling and simulation | |
| CN116911763A (en) | Flow approval method, device, terminal equipment and storage medium | |
| CN114443294B (en) | Big data service component deployment method, system, terminal and storage medium | |
| CN110083351A (en) | Method and apparatus for generating code | |
| CN116402325A (en) | Automatic business process processing method and device | |
| US9177277B2 (en) | Workflow modeling with worklets and transitions | |
| CN115185434A (en) | Vehicle application scene processing method, device, equipment and storage medium | |
| CN110221952B (en) | Service data processing method and device and service data processing system | |
| CN112256978A (en) | Data processing method, device and medium based on data model | |
| CN113806327A (en) | Database design method and device and related equipment | |
| CN112418796A (en) | Sub-process node activation method and device, electronic equipment and storage medium | |
| CN110618778A (en) | Method and system for automatically generating business data, electronic equipment and computer storage medium | |
| US11750458B1 (en) | Structured network change controls |
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 |