Automatic process data processing system

Abstract

The invention discloses an automatic process data processing system, and relates to the technical field of intelligent customer service processing, and the system comprises a management center which is in communication connection with a user side, a data input node, a data processing node and a reply content making node. Customer service demand information initiated by a user is classified, so that corresponding semantic features are extracted according to different types of customer service demand information, and reply contents corresponding to the semantic features are determined through an information graph according to the extracted semantic features. Therefore, the customized reply content corresponding to the customer service demand information initiated by the user is generated, generation of irrelevant reply content is reduced, and reply efficiency is improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of intelligent customer service processing, and particularly relates to a process data automatic processing system. BACKGROUND

[0002] Under the background of rapid development of the Internet and communication technology, customer service demand is growing explosively, and traditional manual customer service has been difficult to meet the two-way demand of enterprises and users, so intelligent customer service data processing technology emerges as the times require. At the beginning, the intelligent customer service data processing mainly uses simple keyword matching, the system identifies the keywords in the user input text by setting rules, and then finds the corresponding reply from the preset answer library; How to generate more customized content for the customer service demand information of the user, reduce the reply of irrelevant content, and improve the reply efficiency of the intelligent customer service is a problem to be solved, and therefore, the present application provides a process data automatic processing system. SUMMARY

[0003] The application aims to provide a process data automatic processing system.

[0004] The application can be achieved by the following technical scheme: a process data automatic processing system, comprising a management center, wherein the management center is in communication connection with a user end, a data input node, a data processing node and a reply content formulation node; The user end is used for inputting customer service demand information by the user; The data input node is used for extracting features of the customer service demand information input by the user, and obtaining corresponding candidate semantic features; The data processing node is used for performing correlation screening on the obtained candidate semantic features, and performing rationality screening on the customer service demand information input by the user according to the correlation screening result; The reply content formulation node is used for formulating key reply content, integrating the formulated key reply content, and obtaining corresponding customized reply content.

[0005] Further, the user inputs customer service demand information through the user end, and the customer service demand information comprises customer service consultation information and customer service after-sales information; An inquiry label is generated according to the customer service consultation information input by the user, an after-sales label is generated according to the customer service after-sales information input by the user, the inquiry label is associated with the customer service consultation information input by the user, and the after-sales label is associated with the customer service after-sales information input by the user.

[0006] Further, the process of extracting features of the customer service demand information input by the user by the data input node to obtain corresponding candidate semantic features comprises: read the label corresponding to the customer service demand information input by the user, and construct a corresponding blank semantic sequence according to the read label; Among them, the blank semantic sequence corresponding to the consultation label is composed of product segments and demand segments; the blank semantic sequence corresponding to the after-sales label is composed of demand segments and time segments; A standard semantic feature library is set up, and semantic terms associated with each commodity in the commodity list and semantic terms associated with after-sales items are constructed in the standard semantic feature library; When the user inputs customer service consultation information, each semantic feature in the customer service consultation information is extracted; Match the extracted semantic features with the semantic terms associated with each commodity in the commodity list. If any semantic feature matches any semantic term associated with any commodity, bind the corresponding semantic feature to the blank semantic sequence corresponding to the commodity, and mark it as a candidate semantic feature.

[0007] Further, when the user inputs after-sales consultation information, extract each semantic feature associated with the after-sales items corresponding to the user's selected commodity in the after-sales consultation information, and match the extracted semantic features with the semantic terms associated with each after-sales item. If any semantic feature matches the semantic term associated with the after-sales item, bind the corresponding semantic feature to the blank semantic sequence corresponding to the commodity, and mark it as a candidate semantic feature.

[0008] Further, the data processing node associates the obtained semantic features, and the process of reasonably screening the customer service demand information input by the user according to the association screening result includes: When the user inputs customer service consultation information, mark the obtained candidate semantic features respectively, and after each semantic term matching the marked candidate semantic features is summarized, the second division is performed according to the commodities associated with each semantic term, and the semantic feature subsets corresponding to each commodity are obtained; Get the number of candidate semantic features in each semantic feature subset, and select the commodity with the most candidate semantic features as the user consultation target commodity; Mark the candidate semantic features matching the semantic terms corresponding to the user consultation target commodity as associated semantic features; Map the associated semantic features to the demand segment in the blank semantic sequence, and import the user consultation target commodity into the product segment, thereby obtaining the consultation semantic segment.

[0009] Further, when the user input is after-sales consultation information, the obtained candidate semantic features are respectively marked, and each semantic entry matching the marked candidate semantic features is respectively summarized, and then secondary division is performed according to the after-sales project associated with each semantic entry to obtain a semantic feature subset corresponding to each after-sales project, and the candidate semantic features in each semantic feature subset are associated with the corresponding after-sales project; The candidate semantic features in the semantic feature subset associated with the after-sales project are introduced into the demand segment in the blank semantic sequence, and the time when the user inputs the customer service after-sales information is introduced into the time segment to obtain an after-sales semantic segment; The time in the time segment is matched with the after-sales time corresponding to the after-sales project, if the time is within the after-sales time, it means that the corresponding after-sales project is reasonable, otherwise it is not reasonable, and the corresponding after-sales project is marked as an overdue project.

[0010] Further, the process of the reply content formulation node formulating key reply content and integrating the formulated key reply content to obtain the corresponding customized reply content includes: An information database is set, and the information database includes information entries associated with each commodity and corresponding after-sales project; An information graph is constructed based on the information entries in the information database; When the user input is customer service consultation information, the consultation semantic segment is input into the constructed information graph; The user consultation target commodity in the consultation semantic segment is read to determine the secondary consultation knowledge node corresponding to the user consultation target commodity; Each semantic feature in the consultation semantic segment is obtained, and each semantic feature is matched with the information entries in each tertiary consultation knowledge node in the information graph, and the information entries associated with the matched tertiary consultation knowledge points are summarized to obtain a reply entry set; When the user input is customer service after-sales information, the after-sales semantic segment is input into the constructed information graph; According to the commodity selected by the user, the secondary after-sales knowledge node corresponding to the commodity is determined; Each semantic feature in the after-sales semantic segment is obtained, and each semantic feature is matched with the information entries in each tertiary after-sales knowledge node in the information graph, and the information entries associated with the matched tertiary after-sales knowledge points are summarized to obtain a reply entry set; After integrating each information entry in the obtained reply entry set, the corresponding customer service customized reply content is obtained.

[0011] Further, the construction process of the information graph includes: According to the consultation label and the after-sales label, corresponding first consultation knowledge nodes and first after-sales knowledge nodes are constructed; According to each commodity in the commodity list, corresponding second consultation knowledge nodes and second after-sales knowledge nodes are constructed; According to the semantic term associated with each commodity, a third consultation knowledge node associated with the second consultation knowledge node is constructed, each third consultation knowledge node corresponds to a semantic term, and each third consultation knowledge node is associated with a corresponding information term; According to the semantic term associated with each after-sales project, a third after-sales knowledge node associated with the second after-sales knowledge node is constructed, each third after-sales knowledge node corresponds to a semantic term, and each third after-sales knowledge node is associated with a corresponding information term.

[0012] Compared with the prior art, the beneficial effects of the present application are: By classifying the customer service demand information initiated by the user, the corresponding semantic features are extracted according to different types of customer service demand information, and the reply content corresponding to each semantic feature is determined through the information graph according to the extracted semantic features, so as to generate customized reply content corresponding to the customer service demand information initiated by the user, reduce irrelevant reply content generation, and improve reply efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments or the prior art, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art based on these drawings.

[0014] Figure 1 The schematic diagram of the present application. DETAILED DESCRIPTION

[0015] As Figure 1 shown, a flow data automatic processing system includes a management center, the management center is in communication with a user end, a data input node, a data processing node and a reply content formulation node; The user end is used for user to input customer service demand information; The data input node is used for feature extraction of the customer service demand information input by the user, and obtains corresponding candidate semantic features; The data processing node is used for associating the obtained candidate semantic features, and performing rationality screening on the customer service demand information input by the user according to the associating screening result; The reply content formulation node is used for formulating key reply content, and integrating the formulated key reply content to obtain corresponding customized reply content.

[0016] It needs to be further explained that the user inputs customer service demand information through the user terminal, and the customer service demand information includes customer service consultation information and customer service after-sales information. In the specific implementation process, the customer service consultation information refers to the relevant consultation information before the purchase of any goods by the user, and the customer service after-sales information refers to the relevant consultation information initiated by the user according to the purchased goods after the user has purchased at least one good.

[0017] According to the customer service consultation information input by the user, a corresponding consultation label is generated, and according to the customer service after-sales information input by the user, a corresponding after-sales label is generated. The consultation label is associated with the customer service consultation information input by the user, and the after-sales label is associated with the customer service after-sales information input by the user. The user can directly send customer service consultation information through the user terminal, and the user needs to select at least one purchased good to send customer service after-sales information.

[0018] It needs to be further explained that the data input node extracts features from the customer service demand information input by the user, and the process of obtaining corresponding candidate semantic features includes: Read the label corresponding to the customer service demand information input by the user, and construct a corresponding blank semantic sequence according to the read label; Among them, the blank semantic sequence corresponding to the consultation label is composed of product segments and demand segments; the blank semantic sequence corresponding to the after-sales label is composed of demand segments and time segments; Set a standard semantic feature library, and construct semantic terms associated with each good in the good list and semantic terms associated with after-sales items in the standard semantic feature library; When the user inputs customer service consultation information, extract each semantic feature in the customer service consultation information; Match the extracted semantic features with the semantic terms associated with each good in the good list. If any semantic feature matches any semantic term associated with any good, bind the corresponding semantic feature to the blank semantic sequence corresponding to the good, and mark it as a candidate semantic feature. When the user inputs after-sales consultation information, extract each semantic feature associated with the after-sales items corresponding to the goods selected by the user in the after-sales consultation information, and match the extracted semantic features with the semantic terms associated with each after-sales item. If any semantic feature matches any semantic term associated with the after-sales item, bind the corresponding semantic feature to the blank semantic sequence corresponding to the good, and mark it as a candidate semantic feature.

[0019] It needs to be further explained that the data processing node associates the obtained semantic features, and the process of reasonably screening the customer service demand information input by the user according to the association screening result includes: When the user inputs customer service consultation information, the obtained candidate semantic features are respectively marked, and each semantic term matching the marked candidate semantic feature is respectively summarized, and then secondary division is performed according to the goods associated with each semantic term, to obtain a semantic feature subset corresponding to each good; The number of candidate semantic features in each semantic feature subset is obtained, and the good with the most candidate semantic features is selected as the user consultation target good; it needs to be explained that if there are multiple semantic feature subsets with the same and maximum number of candidate semantic features, corresponding intention good options are generated, and the intention good options are sent to the user end, and the selected good is selected by the user as the user consultation target good; The candidate semantic features matching the semantic terms corresponding to the user consultation target good are marked as associated semantic features, and the other candidate semantic features are marked as non-associated semantic features; The associated semantic features are mapped to the demand segment in the blank semantic sequence, and the user consultation target good is introduced into the product segment, so as to obtain the consultation semantic segment.

[0020] When the user inputs customer service consultation information, the obtained candidate semantic features are respectively marked, and each semantic term matching the marked candidate semantic feature is respectively summarized, and then secondary division is performed according to the goods associated with each semantic term, to obtain a semantic feature subset corresponding to each good; The candidate semantic features in the semantic feature subset associated with the after-sales project are introduced into the demand segment in the blank semantic sequence, and the time when the user inputs the customer service after-sales information is introduced into the time segment, to obtain the after-sales semantic segment; The time in the time segment is matched with the after-sales time corresponding to the after-sales project, if the time is within the after-sales time, it means that the corresponding after-sales project is reasonable, otherwise it is not reasonable, then the corresponding after-sales project is marked as an overdue project.

[0021] It needs to be further explained that in the specific implementation process, the reply content node formulates key reply content, and integrates the formulated key reply content to obtain corresponding customized reply content. Set up an information database, which includes information terms associated with each good and corresponding after-sales project; Based on the information terms in the information database, an information graph is constructed; It needs to be further explained that in the specific implementation process, the construction process of the information graph includes: According to the consultation tag and the after-sales tag, a corresponding first consultation knowledge node and a first after-sales knowledge node are constructed; According to each commodity in the commodity list, a corresponding second consultation knowledge node and a second after-sales knowledge node are constructed; According to the semantic term associated with each commodity, a third consultation knowledge node associated with the second consultation knowledge node is constructed, each third consultation knowledge node corresponds to a semantic term, and each third consultation knowledge node is associated with a corresponding information term; According to the semantic term related to each after-sales project, a third after-sales knowledge node associated with the second after-sales knowledge node is constructed, each third after-sales knowledge node corresponds to a semantic term, and each third after-sales knowledge node is associated with a corresponding information term; When the user inputs customer service consultation information, the consultation semantic fragment is input into the constructed information graph; Read the user consultation target commodity in the consultation semantic fragment, determine the second consultation knowledge node corresponding to the user consultation target commodity; Get each semantic feature in the consultation semantic fragment, match each semantic feature with the information term in each third consultation knowledge node in the information graph, and aggregate the information terms associated with the matched third consultation knowledge points to obtain a reply term set; When the user inputs customer service after-sales information, the after-sales semantic fragment is input into the constructed information graph; According to the user-selected commodity, determine the second after-sales knowledge node corresponding to the commodity; Get each semantic feature in the after-sales semantic fragment, match each semantic feature with the information term in each third after-sales knowledge node in the information graph, and aggregate the information terms associated with the matched third after-sales knowledge points to obtain a reply term set, and automatically generate "after-sales project has expired" for the overdue project in the after-sales project. The user can continue to select manual customer service; After integrating each information term in the obtained reply term set, the corresponding customer service customized reply content is obtained.

[0022] The above merely describes the preferred embodiments of the present application, and is not intended to limit the present application in any form. Although the present application has been disclosed with the preferred embodiments as above, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content to make equivalent embodiments with equivalent changes, as long as the changes or modifications do not deviate from the technical solution of the present application. Any modification or equivalent replacement to the above embodiments based on the technical essence of the present application still falls within the scope of the technical solution of the present application.

Claims

1. A process data automation processing system, including a management center, characterized in that: The management center is communicatively connected to the user terminal, the data input node, the data processing node and the reply content formulation node; The user terminal is used for the user to input customer service demand information; The data input node is used to extract features from the customer service demand information input by the user to obtain corresponding semantic features to be selected; The data processing node is used to perform relevance screening on the obtained candidate semantic features, and perform rationality screening on the customer service demand information input by the user based on the relevance screening results; The reply content formulation node is used to formulate key reply content and integrate the formulated key reply content to obtain corresponding customized reply content.

2. A process data automation processing system according to claim 1, characterized in that: The user inputs customer service demand information through the user terminal, and the customer service demand information includes customer service consultation information and customer service after-sales information; Generate a corresponding consultation tag based on the customer service consultation information entered by the user, generate a corresponding after-sales tag based on the customer service after-sales information entered by the user, associate the consultation tag with the customer service consultation information entered by the user, and associate the after-sales tag with the customer service after-sales information entered by the user.

3. A process data automation processing system according to claim 2, characterized in that: The data input node extracts features from the customer service demand information input by the user to obtain corresponding semantic features to be selected, including: Read the tag corresponding to the customer service demand information entered by the user, and construct the corresponding blank semantic sequence based on the read tag; Among them, the blank semantic sequence corresponding to the consultation tag consists of product fragments and demand fragments; the blank semantic sequence corresponding to the after-sales tag consists of demand fragments and time fragments; Setting a standard semantic feature library, and constructing semantic entries associated with each product in the product list and semantic entries associated with after-sales items in the standard semantic feature library; When the user inputs customer service consultation information, the semantic features in the customer service consultation information are extracted; The extracted semantic features are matched with the semantic terms associated with each product in the product list. If any semantic feature matches the semantic term associated with any product, the corresponding semantic feature is bound to the blank semantic sequence corresponding to the product and recorded as the candidate semantic feature.

4. A process data automation processing system according to claim 2, characterized in that: When the user inputs after-sales consulting information, the semantic features associated with the after-sales items corresponding to the product selected by the user in the after-sales consulting information are extracted, and the extracted semantic features are matched with the semantic terms associated with each after-sales item. If any semantic feature matches the semantic term associated with the after-sales item, the corresponding semantic feature is bound to the blank semantic sequence corresponding to the product and recorded as a candidate semantic feature.

5. The process data automation processing system according to claim 3, characterized in that: The data processing node performs relevance screening on the obtained semantic features, and performs rationality screening on the customer service demand information input by the user based on the relevance screening results, including: When the user inputs customer service consultation information, the obtained candidate semantic features are marked respectively, and the semantic terms that match the marked candidate semantic features are summarized respectively. Then, a secondary division is performed according to the products associated with each semantic term to obtain the semantic feature subset corresponding to each product; Obtain the number of candidate semantic features in each semantic feature subset, and select the product with the largest number of candidate semantic features as the user's inquiry target product; Marking the candidate semantic features that match the semantic terms corresponding to the target product consulted by the user as relevant semantic features; The associated semantic features are mapped to the demand segment in the blank semantic sequence, and the user's inquiry target product is imported into the product segment to obtain the inquiry semantic segment.

6. A process data automation processing system according to claim 4, characterized in that: When the user inputs after-sales consultation information, the obtained candidate semantic features are marked respectively, and the semantic terms that match the marked candidate semantic features are summarized respectively. Then, secondary division is performed according to the after-sales items associated with each semantic term to obtain semantic feature subsets corresponding to each after-sales item. The candidate semantic features in each semantic feature subset are then associated with the corresponding after-sales item. Import the selected semantic features in the semantic feature subset associated with the after-sales item into the demand segment in the blank semantic sequence, and import the time when the user inputs the customer service after-sales information into the time segment to obtain the after-sales semantic segment; Match the time in the time segment with the after-sales time corresponding to the after-sales project. If the time is within the after-sales time, it means that the corresponding after-sales project is reasonable. Otherwise, it is unreasonable, and the corresponding after-sales project will be marked as an overdue project.

7. A process data automation processing system according to claim 6, characterized in that: The reply content formulation node formulates key reply content and integrates the formulated key reply content to obtain corresponding customized reply content. The process includes: Setting up an information database, wherein the information database includes information entries associated with each product and the corresponding after-sales items; Construct corresponding information graphs based on information entries in the information database; When the user inputs customer service consultation information, the consultation semantic fragment is input into the constructed information graph; Read the user's inquiry target product in the inquiry semantic segment and determine the secondary inquiry knowledge node corresponding to the user's inquiry target product; Obtain each semantic feature in the consultation semantic segment, match each semantic feature with the information terms in each third-level consultation knowledge node in the information graph, and aggregate the information terms associated with the matched third-level consultation knowledge points to obtain a response term set; When the user inputs customer service after-sales information, the after-sales semantic fragment is input into the constructed information graph; According to the product selected by the user, determine the secondary after-sales knowledge node corresponding to the product; Obtain each semantic feature in the after-sales semantic segment, match each semantic feature with the information terms in each third-level after-sales knowledge node in the information graph, and aggregate the information terms associated with the matching third-level after-sales knowledge points to obtain a response term set; After integrating the various information entries in the obtained reply entry set, the corresponding customer service customized reply content is obtained.

8. A process data automation processing system according to claim 7, characterized in that: The process of constructing the information graph includes: According to the consulting label and after-sales label, the corresponding first-level consulting knowledge node and first-level after-sales knowledge node are constructed respectively; Build corresponding secondary consulting knowledge nodes and secondary after-sales knowledge nodes for each product in the product list; Constructing third-level consulting knowledge nodes associated with the second-level consulting knowledge nodes based on the semantic terms associated with each product, each third-level consulting knowledge node corresponds to a semantic term, and associating corresponding information terms with each third-level consulting knowledge node; According to the semantic terms related to each after-sales project, the third-level after-sales knowledge nodes associated with the second-level after-sales knowledge nodes are constructed. Each third-level after-sales knowledge node corresponds to a semantic term, and each third-level after-sales knowledge node is associated with the corresponding information term.

Images