TWI646528B - System for conducting dialogues - Google Patents

Abstract

This case involves a system for continuing and guiding conversations. The communication between the first server and the electronic device is mainly provided to provide a dialogue interface and obtain feedback information. The second server stores conversation information, a conversation feature set, and associated information. The third server is connected to the first server and the second server, and can follow a decision tree or according to the feedback information to select a first statement or a second statement in the dialog information. The fourth server is connected to the server, and the semantic analysis is performed when the feedback information cannot correspond to the decision tree, and a third statement is selected according to the analysis result and the related information. The first, second, and third statements will be combined with the conversation feature set to provide to the conversation interface.

Description

Dialogue guidance system

The case relates to an application system, in particular a system for continuing and guiding conversations.

Along with the popularity of digital technology, industries inevitably have to extend or transfer their services to various electronic devices. However, the service nature of certain industries focuses on the communication between people. If the general service or the hard-to-read data is provided only through mobile devices, the service status is inevitably less than ideal. It may be possible to provide a more complete service by using customer service staff to conduct text or voice conversations with customers on the Internet. However, such a service that imposes a burden on actual human resources is not desirable for the industry. Therefore, how to provide an efficient automated customer service system and provide customers with a feeling of being a real person in service or promotion is an important research topic in the field.

One of the objectives of this case is to provide a dialogue guidance system to improve the interactivity and poor user experience of prior art automated customer service systems.

The present disclosure discloses a dialog guidance system, which includes: a first server, the first server is communicatively coupled to an electronic device to provide a dialog interface for displaying on the electronic device, and obtaining a user feedback from the dialog interface. a second server for storing at least one dialog information, at least one dialog feature set or at least one associated information; a third server, the third server being respectively connected to the first server and the second server And the third server follows the decision tree to select at least one first statement in the at least one conversation information, or searches the feedback tree for the feedback information to select at least one second statement in the at least one conversation information, and at least one a statement or at least one second statement is provided to the dialog interface in combination with the at least one dialog feature set; and a fourth server is respectively connected to the first server, the second server, and the third server, wherein the feedback information is When the decision tree is not available, the fourth server performs a semantic analysis on the feedback information to select at least one pair according to the result of the semantic analysis and the at least one associated information. Information at least a third statement, then at least a third statement in conjunction with the feature set provided at least a dialogue to dialogue interface.

Therefore, according to the technical content of the present case, the embodiment of the present invention improves the interaction and user experience of the customer service system by providing a dialogue guidance system.

The spirit of the present invention will be clearly described in the following drawings and detailed descriptions. Anyone having ordinary knowledge in the technical field will be able to change and modify the technology as taught by the present invention after learning the embodiments of the present invention, and does not deviate from the present case. Spirit and scope.

The terminology used herein is for the purpose of describing particular embodiments and is not intended to The singular forms "a", "the", "the", "the" and "the" are also used in the plural.

The use of the terms "first", "second", ", etc." as used herein does not specifically mean the order or the order, and is not intended to limit the present invention. It is merely to distinguish between elements or operations described in the same technical terms.

As used herein, "coupled" or "connected" may mean that two or more elements or devices are in direct physical contact with each other, or indirectly in physical contact with each other, or two or more elements or devices. Operation or action.

The terms "including", "including", "having", "containing", etc., as used in this document are all open terms, meaning, but not limited to.

With respect to "and/or" as used herein, it is meant to include any or all combinations of the recited.

Regarding the directional terms used in this article, such as: up, down, left, right, front or back, etc., only refer to the direction of the additional schema. Therefore, the directional terminology used is used to illustrate that it is not intended to limit the case.

The terms used in this document, unless otherwise noted, usually have the usual meaning of each term used in this field, in the context of the case, and in particular content. Certain terms used to describe the present invention are discussed below or elsewhere in this specification to provide additional guidance to those skilled in the art in the description of the present disclosure.

FIG. 1 is a schematic diagram of a dialog guiding system according to an embodiment of the present invention. In one embodiment, the dialog guidance system 100 includes a web server 110, a conversation guide server 120, a database server 130, and a dialog analysis server 140. Network server 110 can be any existing or feasible communication server that is used to connect to the network to send or receive information. The dialog guide server 120 can be any existing or feasible option for executing a program or operation to produce a processing result. The database server 130 can also be any existing or feasible storage facility, such as a read only memory (ROM), a hierarchical memory or a database management server, etc., and is mainly used for storing information for a long time. . The dialog analysis server 140 can be any existing or feasible option, which is also used to execute a program or operation to produce a processing result.

In this embodiment, an external electronic device 200 is further illustrated. The electronic device 200 can be any existing or feasible computer device, such as a smart phone, a tablet computer, a desktop computer, or the like. In general, the electronic device 200 has a display screen to display program execution results or received information for the user to understand.

In this embodiment, the electronic device 200 is communicatively coupled to the dialog guidance system 100 via the network, and the information transmitted by the electronic device 200 is first received by the network server 110 and then transmitted to the conversation guide server 120 or the conversation. The server 140 is analyzed. In contrast, the information sent by the dialog guide server 120 or the dialog analysis server 140 will also be transmitted to the network via the network server 110 and then transmitted to the electronic device 200. In other words, the network server 110 is a conduit for the dialog guiding system 100 to communicate with external devices. The network server 110 may include a unit for implementing message encryption and decryption or other network management mechanisms to maintain the dialog guiding system. 100 information security.

In this embodiment, the database server 130 stores conversation information, a conversation feature set, associated information, or product information. The dialogue information includes questions and answers in human daily life or professional activities, question and answer sequence, auxiliary words, situational terms, professional words, etc. The dialogue information may also include specific information extracted from other programs or network information. For example: popular terms, subculture terms, weather forecasts, news content, etc. The dialog feature set is a combination of a set of dialog feature sets that can be used to present or construct a vocal form or habit of a virtual character, and the avatars shaped by the set of dialog features each have substantially different genders, special terms, Emotion, tone, mantra, sales weight, and dialogue extension settings, for example, a set of dialogue features may be used to present the expression of a friendly and young girl, while another set of dialogue features is used Presenting the attitude of an old and active older man, and the setting of the proportion of sales can determine the type and frequency of virtual personality marketing.

In this embodiment, the associated information in the database server 130 is used to indicate the association between a specific statement and a dependent statement in a human natural language. For example, the continuation statement of the "hello" statement may have "you". Good, "嗨" or "true". The correlation between a specific statement and a complex subordinate statement can be matched with the respective weights. For example, in the above example, "true" may be a continuation statement of the "hello" statement. However, this usage is less. Commonly, the weight of the relevance of "Hello" to "Hello" may be lower than the relevance of "嗨" to "Hello." The product information in the database server 130 includes various services of the industry and prices, plans, purchase qualifications, conditions, detailed data, trial balances and the like of various products.

In this embodiment, a dialog interface is displayed on the electronic device 200. The dialog interface is provided to the electronic device 200 via the network server 110. The dialog interface is used to provide a user interactive service window. The dialog guidance system 100 can present information to the user through the dialog interface, and the dialog guidance system 100 can obtain feedback information generated by the user interacting with the dialogue interface. . This dialog interface can be used for any existing or viable options, such as a dialog window similar to instant messaging or online customer service, plus a number of menus or features that enhance the user experience.

In this embodiment, the dialog guidance server 120 is connected to the network server 110 and the database server 130 respectively to follow a decision tree to select at least one first statement in the dialog information, or to search in the decision tree. The feedback information is used to select at least one second statement in the conversation information, and then the at least one first statement or the at least one second statement is provided to the conversation interface.

In the present embodiment, the dialog analysis server 140 is connected to the web server 110, the dialog guide server 120, and the database server 130, respectively. When the feedback information of the user cannot correspond to the decision tree, the dialog analysis server 140 performs semantic analysis on the feedback information, and selects at least one third statement in the dialog information according to the semantic analysis result and the associated information, and then provides the third statement. To the dialogue interface. In other words, the dialog analysis server 140 can perform a semantic analysis on the feedback information, and then select possible response statements that are not in the decision tree according to the correlation or the dependency between the known sets of statements. In addition, after the conversation analysis server 140 performs semantic analysis, it may also return the analysis result to the conversation guidance server 120 to search the decision tree, and then select a fourth statement in the dialog information, and then provide the fourth statement to the conversation. interface.

The decision tree of the present case includes a plurality of nodes, and each node includes a plurality of branches, and each node and its branches are developed according to the semantic subordination and relevance between the dialogue information of the present case. A node of the decision tree corresponds to at least one of the dialog information, and each branch under the node corresponds to at least one of the dialog information. When the dialog interface is initialized, the dialog guidance server 120 may select the dialog information corresponding to a node as the first statement, and the node is usually the root of the decision tree, and if the last dialogue in the dialog interface is to be continued, The decision tree searches for a node similar to the last conversation to select the dialog information corresponding to the branch under the node. If the dialog information corresponding to a node matches or is similar to the user's feedback information, the dialog guidance server 120 may select the dialog information corresponding to one of the branches under the node as at least one second statement.

Please refer to FIG. 2, which is a schematic diagram of a decision tree according to an embodiment of the present invention. Its main structure is a binary tree, and each node contains multiple dialog information to choose from. Among them, the root node R of this decision tree is a kind of "good question" type, and the dialogues include "good morning", "good afternoon", "good night", "嗨" and other dialogue information, and the "good question" may be followed. The type of "Respond to Hello" or the type of "Precaution Statement" will be described as follows. See the left branch L1 of the root node, which is a "response hello statement", including dialog information such as "you / 妳 early", "good morning", "ha 啰". See the right branch L2 of the root node, which is a "precaution statement", including dialog information such as "what is wrong" and "somewhat busy". In addition, the left sub-branch L11 of the "Respond to Hello" type may have a "Open Topic" category, which includes dialogue information such as "Good Weather," "How Mood," and "Needs Service". The right branch L21 of the type of "precautionary question" may be the "avoidance" type, including "no problem" and "nothing". The dialog guiding server 120 of the present case searches for the feedback information of the user in the form of a decision tree like this, and then selects the dialogue information for responding to the feedback information.

In an embodiment of the present invention, in order to adapt to different languages or situations, more than one corresponding decision tree may be provided to facilitate the system to perform sentence search and selection. In addition, since various statements may have commonality in different situations, different types of nodes may contain the same conversation information in the decision tree of the case. For example, "Good Morning" can be a kind of active greeting or a good response. The statement, so it can be included in several nodes. The system of this case will judge whether to select the same statement in different nodes of the decision tree through the order of the dialogue. For example, the feedback information of the user in the previous dialogue includes "Good Morning", and the system selects "Good Morning" to respond to the question. Although the response message belongs to the same statement, the two are actually "good morning" statements in nodes of different levels of the decision tree.

In the second embodiment, in this embodiment, the dialog guidance server 120 first selects "Good Morning" in the root node R "Well Statement" as the first sentence to provide to the dialog interface to try to ask the user. Subsequently, the user briefly responds to "Hello" to the conversation interface, and the user's response "Hello" is transmitted as feedback information via the web server 110 to the conversation guide server 120, and the dialog guidance server 120 After finding the decision tree, it is found that the feedback information can correspond to the "response hello statement" of the left branch L1, and then the dialog guide server 120 selects "how to be in the "open topic" category of the left branch L11 below the left branch L1. As the second statement, try to respond to the user with the second statement.

However, it is worth noting that the dialog guidance server 120 of the present invention can not only provide the first statement or the second statement to the dialog interface, but also can combine the dialog feature set to present the first statement or the second statement to the dialog interface. As described above, since the dialog feature set of the present case can be used to present or construct a dialogue character or habit of a virtual character, each virtual character created by each dialog feature set may have different genders, special terms, mantras, etc. If the first statement or the second statement is combined with the dialog feature set to be presented in the dialog interface, each statement can be represented by the character feature of the dialog feature set. Through the characteristics of the characters shaped by the dialogue feature set, the degree of similarity between the system and the real person can be improved, and the user's defense mentality can be lowered to establish a trust relationship.

In summary, an embodiment of the present invention provides a dialog guidance system to automatically present selected statements to a dialog interface through a dialog feature set, thereby improving the prior art service interactivity and poor user experience.

Please refer to FIG. 3, which is a first schematic diagram of a dialog interface according to an embodiment of the present invention. In this embodiment, each statement will be presented through a first dialog feature set for presenting a avatar of a character having the following personality (name: Xiaotao, gender: female, personality: cheerful, special) Terminology: Like to use popular words and emoji, topic extension: use food and its own state as an extension). In the embodiment of FIG. 2, the dialog guidance server 120 first selects "Good Morning" in the root node R as the first statement. According to the above setting, the first sentence will construct the feature format through the first dialog feature set. It is decorated or expressed in the dialogue interface and is displayed as "You are good morning:).

In this embodiment, in order to facilitate the user's response, the dialog guidance system 100 will provide the user with a convenient response option 301 and a response option 302, and the system will divide the left node of the root node R of the "good question" from the decision tree. Each of the "Response to Good Question" type in the branch L1 and the "Politeness Question" type in the right branch L2 are selected in the dialogue interface in the form of a response option. The dialog guidance system 100 selects "Yes" as the default response option 301, and "a bit busy" as the default response option 302, and then provides the response options 301 and 302 to the dialog interface for the user to click. If the user clicks on the response option, it is regarded as a kind of feedback information transmission to the dialogue guiding system 100. In addition, it is considered that the user may not be able to select the sentences that actually want to respond in response to the options 301 and 302. The dialog guiding system 100 further provides a free input unit 303 to the dialog interface for the user to directly input text or symbols and input the free input unit. The information of 303 will be treated as a kind of feedback information transmitted to the conversation guidance system 100.

Please refer to FIG. 4 again, which is a second schematic diagram of the dialog interface of an embodiment of the present invention. Following the embodiment of FIG. 3, in this scenario, the user selects the response option 301 in FIG. 3 in the dialog interface, and the response option 301 "妳好" will be sent back to the conversation guidance system 100 as feedback information. The bootstrap server 120 then looks up the decision tree and confirms that there is a node that matches the semantics of "good", which will select "how the mood" in the branch under the node as the second statement. The second sentence is expressed in the dialogue interface by the feature format constructed by the first dialog feature set, and the second sentence is added to the dialog set extension of the first feature set to become "Little Peach has a delicious burger for breakfast today. There is a feeling of satisfaction. What is your mood today?". In this way, the system no longer presents a flat single sentence, but a human-like dialogue with a more three-dimensional personality, which will help to enhance the user experience. In this embodiment, similarly, the dialog guidance system 100 will also provide the user with a corresponding response option, the response option 301 will be replaced by "very good", and the response option 302 will be replaced by "not very good". Users quickly respond.

Please refer to FIG. 5 again, which is a third schematic diagram of the dialog interface of an embodiment of the present invention. Following the embodiment of Fig. 4, in this scenario, a number of person tags are provided on the interface side of the dialog, and each character tag provides an option for the user to switch between different dialog feature sets. In other words, the user can change the dialog feature set by switching the character tag, so that the dialog presented by the system in the dialog interface seems to be provided by other virtual characters. In this embodiment, the user does not like the speech of the peach, so the character tag above the dialog interface is switched. Thereafter, each statement is presented through a second dialog feature set, and the second dialog feature set is used for rendering. The character of the avatar with the following personality (name: Azheng, gender: male, personality: serious, special language: honorific, topic extension: service and merchandising sales as an extension).

In the embodiment of FIG. 5, the dialog guide server 120 first selects the root node "Good Morning" as the first statement. According to the above setting, the first sentence will be modified by the feature format constructed by the second dialog feature set or Expressed in the dialogue interface, it is displayed as "You are early, please call me Ah Zheng". Similarly, at a later time in the context depicted in this figure, the dialog guidance system 100 may select "Hello" and "Slightly Busy" as the default response options 301 and 302 for the user to click. The dialog guidance system 100 will likewise provide a free input unit 303 to the dialog interface for the user to freely enter text or symbols. At this point in time, the user clicks on the response option 301 in the dialog interface, and the "hello" in the response option 301 is sent back to the conversation guidance system 100 as feedback information.

In this embodiment, the dialog guidance server 120 then looks up the decision tree and confirms that there is a matching node, which will select "need to serve" in the branch under the node as the second statement. The second statement is modified or expressed in the dialog interface by the feature format constructed by the second dialog feature set, because the dialog extension setting of the second dialog feature set includes the merchandise sales, when providing the dialogue to the conversation interface, A number of product information is called from the database server 130, and the second statement and the product information pass the feature format of the second feature set and become "very happy to serve you. The company just launched a new ETF stock this week, may I ask you? Interested in understanding?", that is, the immediate situation of the situation is depicted in this diagram. In this manner, the dialog guidance system 100 can present a humanoid conversation with a more stereoscopic personality. In this embodiment, the dialog guiding system 100 can provide the user with the "I am interested" or "not interested" two-sentence option as the default response option 301 and the response option 302, so that the user can quickly respond. . The dialog guiding system 100 further provides a free input unit 303 to the dialog interface for the user to directly input text or symbols, and the information input to the free input unit 303 is regarded as a kind of transmission information to the dialog guiding system 100.

Please refer to FIG. 6 , which is a fourth schematic diagram of a dialog interface according to an embodiment of the present invention. According to the figure, in an embodiment, after a certain dialogue between the user and the system in the dialogue interface, the user inputs freely. In unit 303, enter "I want to help my mother find a long-term insurance plan." And "I want to help my mother find a long-term insurance plan" this statement is of course regarded as a kind of feedback information, which will be transmitted to the dialogue guiding system 100. However, this statement is not similar to any dialog category of the decision tree of the dialog guide server 120, so the dialog guide server 120 cannot find the corresponding dialog information. At this time, the dialogue analysis server 140 will analyze the semantics of "I want to help my mother find a long-term insurance plan". For example, the statement can be disassembled into "I", "Mom", "Long Life Insurance" and other key points. word. The "I" statement can return to the decision tree of the dialog guide server 120 to find the type of "respond to the user's self-proclaimed", so that the system selects the fourth sentence "you" to respond. The "Mom" statement may find the third sentence "Langtang" and "Home" related to it in the related information. The "Long Life Insurance" statement may find information such as "Long Photo Protection Cancer 2.0" related to the product information of the database server 130.

According to the above embodiment, the dialog guidance system 100 comprehensively judges according to the third sentence, the fourth sentence, and the product data obtained above, and the appropriate semantics of the user feedback information, and then presents the conversation to the conversation through the currently selected conversation feature set. interface. For example, in the above embodiment, when the dialog analysis server 140 matches the "mother" statement with the related information, the two related third sentences "Yuantang" and "home" are found. However, "Langtang" and "home" are found. The relevance of the two statements to the "mother" statement has its own weight value. In this embodiment, the subject sentence of "Mom" first is "I", so the system combines the semantics of the two sentences to judge that "Langtang" has a higher weight, so the statement is selected. In addition, since the "order" in the "Lingtang" is synonymous with the fourth sentence "you", it is not necessary to add the "you" sentence in the final response sentence to remove the word.

According to the above example, corresponding to the user's input "I want to help my mother find a long-term insurance plan", the dialogue analysis server 140 performs semantic analysis and selects the response sentence, which may be "no problem! Our company launched a new service last week." Long-term anti-cancer 2.0, would you like to ask the court to live this year? I will check if the age of the court is in line with the conditions of this program." This statement is provided to the dialogue interface. Similarly, the dialog guidance system 100 selects "wait a moment" and "has a choice" as the default response options 301 and 302 for the user to respond by clicking. It can be seen that the dialogue guided by the system of this case is obviously more human and efficient than the previous technology.

It can be seen from the above embodiments of the present invention that the application of the present invention has the following advantages. The embodiment of the present invention provides a dialogue guiding system, provides a dialogue interface through a combination of a decision tree and various dialog feature sets, so that the user feels a conversation more similar to a real person, and improves the problem of poor service interactivity in the past. The dialog guidance system of the embodiment of the present invention further has a dialog analysis function. When the feedback of the user cannot correspond to the decision tree, the feedback content of the user is analyzed to obtain a statement for answering. The embodiment of the present invention provides a response option to the dialog interface, so that the user can directly respond in a selected manner. The embodiment of the present invention provides a switching of the feature set of the dialogue interface to increase the degree of freedom of the user in the service environment.

Although the present invention is disclosed in the above embodiments, it is not intended to limit the present invention. Anyone skilled in the art can make various changes and refinements without departing from the spirit and scope of the present invention. Therefore, the scope of protection of this case is attached. The scope of the patent application is subject to change.

100: dialog guidance system 110: network server 120: dialog guidance server 130: database server 140: dialog analysis server 200: electronic device 301: response option 302: response option 303: free input unit R: Root node L1: Left branch L2: Right branch L11: Left branch L21: Right branch

1 is a schematic diagram of a dialog guidance system according to an embodiment of the present invention; FIG. 2 is a schematic diagram of a decision tree according to an embodiment of the present invention; FIG. 3 is a first schematic diagram of a dialog interface according to an embodiment of the present invention; The second schematic diagram of the dialog interface of the example; FIG. 5 is a third schematic diagram of the dialog interface of an embodiment of the present invention; and FIG. 6 is a fourth schematic diagram of the dialog interface of an embodiment of the present invention.

Claims (8)

A dialog guiding system includes: a first server, communicatively coupled to an electronic device, providing a dialog interface displayed on the electronic device, and obtaining a feedback information from the dialogue interface; and a second server storing at least one The dialog information, the at least one dialog feature set, and the at least one associated information; a third server connected to the first server and the second server respectively, and selecting at least one of the at least one dialog information according to a decision tree a statement, or searching the decision tree according to the feedback information to select at least one second statement in the at least one dialog information; and a fourth server connecting the first server, the second server, and the third a server, when the feedback information cannot correspond to the decision tree, the fourth server performs a semantic analysis on the feedback information, and the feedback information is disassembled into at least one keyword, and the at least one associated information is selected according to the analysis result and the at least one associated information. At least one third statement associated with the at least one keyword in the dialog information; wherein the first server is to perform the at least one first statement, the at least one The statement or the at least one third statement will be combined with the at least one conversation feature set to provide to the conversation interface, wherein the at least one conversation feature set each includes a different gender setting, mood setting, special language, tone, mantra, marketing a weight or dialog extension setting, wherein at least two of the at least one first statement, the at least one second statement, the at least one third statement, or the at least one fourth statement are provided for the conversation interface The response statement is also provided in the dialog interface for the user to select. When the user selects one of the at least two response statements to respond, one of the response statements is regarded as one of the feedback information. The system of claim 1, wherein the node and the branch of the decision tree are developed according to semantic preference and relevance between the at least one dialog information, and each node of the decision tree and each branch under the node Each of the at least one conversation information corresponding to the at least one conversation information is corresponding to the feedback information, and the third server selects the at least one conversation information corresponding to one of the branches under the node as The at least one second statement. The system of claim 1, wherein the second server further stores at least one product information, wherein the feedback information, the at least one first statement, the at least one second statement, or the at least one third statement relates to the at least one The product information is combined with the at least one conversation feature set to provide to the conversation interface. The system of claim 1, wherein the at least one associated information is used to represent an association between a particular statement and a dependent statement in the natural language, and the association is represented by at least one weight. The system of claim 1, wherein after the semantic analysis of the fourth server, the statement relevance involved in the semantic analysis is stored in the at least one associated information. The system of claim 1, wherein after the semantic analysis is performed by the fourth server, the analysis result is returned to the third server to search the decision tree, and at least one of the at least one dialog information is selected. The statement combines the at least one fourth statement with the at least one conversation feature set to provide to the conversation interface. The system of claim 1, wherein the first server further provides a free input unit to the dialog interface for the user to input text or symbols, and the information input to the free input unit is regarded as one of the feedback information. The system of claim 1, wherein the dialog interface provides an option to switch the at least one conversation feature set.