KR20200092448A - AI interactive phone ordering system for T commerce - Google Patents

Abstract

The present invention provides an artificial intelligence interactive home shopping phone ordering method and a system thereof. According to an embodiment of the present invention, the artificial intelligence interactive home shopping phone ordering method comprises: a step in which a legacy server extracts product information of a product which is currently in a home shopping broadcast; a step in which an interactive voice response server receives a voice input of a customer; a step in which a speech-to-text (STT) processing module converts the voice input into text information; and a step in which an artificial intelligence application performs an ordering procedure of the product based on the converted text information in connection with the legacy server. Accordingly, the ordering service effect can be maximized even for customers who are not familiar with ARS automatic ordering and prefer ordering with service associates by natural language sentence conversation phone ordering similar to a conversation with an ordinary service associate instead of button-type ARS or short sentence speech recognition. Moreover, all content for ordering service content is converted into text to store the text and the changing trend of interests of customers and issue items can be identified.

Description

AI interactive phone ordering system for T commerce}

The present invention relates to an artificial intelligence interactive home shopping telephone ordering method and system, and more specifically, an artificial intelligence interactive home shopping telephone ordering method and system utilizing STT (Speech to Text) technology, artificial intelligence NLC (natural language analysis) technology. It is about.

In a home shopping broadcast, the most important thing for ordering a product that is on sale is to respond to a product inquiry to induce an order to a customer and to easily access a customer who prefers to order a counselor, but when a phone is connected to purchase a product being home shopping, product inquiry There is a lot of complaints about the waiting of consumers because the wait time for the counselor for the service is long, and in the case of automatic order, only the work related to the order is processed, and the automatic order listens to the announcement and proceeds with the purchasing process with the key input of the phone This is uncomfortable.

In order to overcome this, a voice-based automatic order service can be used, but this is a short-answer short-talk voice recognition, and the use area of the voice-based is very small. This exists.

The present invention has been devised to solve the above problems, and the purpose of the present invention is to use artificial intelligence to grasp the intention of natural language-based sentences rather than short conversation recognition services and place product orders with appropriate answers and feedback. It is to provide a voice-based artificial intelligence interactive home shopping phone ordering method and system beyond the function of simple automatic ordering by induction.

In addition, another object of the present invention is to provide an artificial intelligence interactive home shopping telephone ordering method and system that can store and store all contents in text for order consultation contents, and grasp changes in customer interests and issues. .

In accordance with an embodiment of the present invention for achieving the above object, extracting product information of the product currently being broadcast home shopping; Receiving a voice input from a customer; Converting the voice input into text information; And performing an order procedure for the product based on the converted text information.

In addition, the step of extracting the product information includes: storing mapping information generated by mapping product information of a plurality of products and home shopping broadcast schedule information of each product; And extracting product information of a product currently being broadcast on home shopping among the stored mapping information.

And extracting the product information, after the product information is extracted, comparing the schedule information and the current time of the extracted product information, verifying the validity; And when validation fails, determining that the product information is not suitable for the current time, and extracting the product information again.

The step of performing the ordering process of the product may determine whether an event occurs in the process of performing the ordering process based on the converted text information, so that when the event occurs, the customer is connected to the telephone agent.

In this case, the event may be a situation in which the unstructured content included in the converted text information is judged to be unsuitable for text analysis using natural language processing, and the order procedure is not normally performed.

In addition, the event extracts sentiment as metadata through text analysis from the unstructured content included in the converted text information, and analyzes the emotion of a specific phrase and the emotion of the entire document, and the customer's emotional state is dissatisfied. It can be a situation judged to be one of, annoyance, and anger.

And, when the member information on the age of the customer is obtained, the event occurrence threshold may be set according to the age of the customer so that the event is easily generated as the age of the customer increases.

In addition, the step of performing the ordering process includes: obtaining member information of the customer; Converting greetings and introductions related to product purchase into voice information; Collecting purchase information including purchase options, quantities and addresses of products; And performing a payment procedure.

And the step of collecting the purchase information, inferring the address information of the upper layer based on the address information of the lower layer having an accuracy of a threshold or higher in the logical hierarchical structure of the address when the text analysis of the address fails. Can.

In addition, at this time, the logical hierarchical structure of the address information includes the first tier including the city/county/province/city/city/county/eup/myeon, and the second tier as the lower tier of the first tier according to the administrative district. , Dong/ri is included, the third layer, which is a lower layer of the second layer, and the barrel/ban/address/lake, may be composed of the fourth layer, which is the lower layer of the third layer.

On the other hand, according to another embodiment of the present invention, the artificial intelligence interactive home shopping telephone ordering system, a legacy (legacy) server for extracting product information of the product currently being broadcast home shopping; An interactive voice response server that receives a customer's voice input; A speech to text (STT) processing module that converts speech input into text information; And an artificial intelligence application that performs an order procedure for a product based on the converted text information in cooperation with a legacy server.

Meanwhile, according to another embodiment of the present invention, the artificial intelligence interactive home shopping telephone order server includes a storage unit in which product information of a product is stored; An interactive voice response processing unit that receives a customer's voice input; A speech to text (STT) processing unit that converts voice input into text information; It includes; and a processor for extracting the product information of the product currently being broadcast home shopping among the stored product information, and performs an order procedure for the product based on the converted text information.

As described above, according to the embodiments of the present invention, customers who are not familiar with ARS automatic ordering of a preferred order of a counselor through a conversational phone order of a natural language sentence similar to a conversation with a general counselor rather than button-type ARS or short speech recognition. You can also maximize the effectiveness of order consultation.

In addition, according to embodiments of the present invention, it is possible to grasp and store all contents in text for order consultation contents, and to grasp changes in customer interests and issues.

1 is a diagram provided to explain the configuration of an artificial intelligence interactive home shopping telephone ordering system according to an embodiment of the present invention;
Figure 2 is a diagram provided in the description of the artificial intelligence interactive home shopping phone ordering method according to an embodiment of the present invention,
3 is a diagram provided for explaining the ordering process of a product by the artificial intelligence interactive home shopping telephone ordering method according to an embodiment of the present invention, and
4 is a diagram provided to explain the configuration of an artificial intelligence interactive home shopping telephone order according to another embodiment of the present invention.

Hereinafter, the present invention will be described in more detail with reference to the drawings.

1. Artificial intelligence interactive home shopping phone ordering system

In an embodiment of the present invention, an artificial intelligence interactive home shopping telephone ordering system and method are presented.

"Artificial intelligence interactive home shopping phone ordering system" uses artificial intelligence to understand the intention of natural language-based sentences rather than short conversational recognition services and induce product orders with appropriate answers and feedback to order beyond the functions of simple automatic ordering It is a system that provides counseling services.

1 is a view provided to explain the configuration of an artificial intelligence interactive home shopping telephone ordering system according to an embodiment of the present invention. As shown in FIG. 1, the artificial intelligence interactive home shopping telephone ordering system according to an embodiment of the present invention includes an interactive voice response server 100, a speech to text (STT) processing module 200, It may be composed of an artificial intelligence application 300, a legacy server 400, a text to speech (TTS) processing module 500 and an associated server 600.

When the telephone is connected from the customer terminal 10, such as a telephone or a smart phone, the interactive voice response server 100 may receive the voice input of the customer, and may transmit voice information in response to the voice of the customer.

Specifically, the interactive voice response server 100 may accept a combination of voice input from a customer and selectively pressing a key pad of the customer terminal 10 to provide an appropriate response in the form of voice, callback, or other media. have.

The STT processing module 200 may convert voice input into text information and transmit it to the artificial intelligence application 300 and the legacy server 400.

The artificial intelligence application 300 may work with the legacy server 400 to provide appropriate answers and feedback to the customer's questions and requirements based on the converted text information, and perform a product ordering process.

The artificial intelligence application 300 may use information stored in the legacy server 400 in conjunction with the legacy server 400 and access the artificial intelligence server 20 to use artificial intelligence algorithms.

Specifically, the artificial intelligence application 300, through the STT processing module 200, through the text analysis on the unstructured content included in the converted text information, analyzes and recognizes the customer's speech content, grasps the customer's intention And, check the product information, member information, etc. stored in the legacy server 400, and may perform an order procedure.

To this end, the artificial intelligence application 300 analyzes text from the unstructured content included in the text information converted using the natural language understanding algorithm, and is not directly specified in categories and text information that classify content using a plurality of classification layers. Concepts that identify unconceptual concepts, keywords that identify related keywords based on content, analyze sentences in the form of subjects, activities, and targets, semantic roles that identify the subject or objects of a particular activity, or keywords. Metadata such as sentiment for analyzing sentiment and sentiment for the whole document, and relations for identifying the relationship type between two entities can be extracted.

In addition, the artificial intelligence application 300 determines whether an event occurs for a situation in which an order procedure is not normally performed in the process of ordering a product based on the converted text information, and when an event occurs, a customer is a telephone counselor It is possible to minimize the customer's inconvenience by making a phone call to.

The legacy server 400 stores the mapping information generated by mapping the product information of a plurality of products and the home shopping broadcast schedule information of each product, extracts product information of the product currently being broadcast on the home shopping, and artificial intelligence application 300 It is possible to perform the ordering process for this product.

Also, after the product information is extracted, the legacy server 400 compares the schedule information of the extracted product information with the current time, verifies the validity, and when validation fails, the product information is not suitable for the current time. By judging, product information can be extracted again.

The TTS processing module 500 may convert the written text information into voice information implemented in various languages and voices, and the associated server 600 includes an interactive voice response server 100 and an STT processing module 200 and TTS processing It is provided to link the transmission and reception of information between the modules 500.

2. How to order an artificial intelligence interactive home shopping phone

FIG. 2 is a diagram provided for explaining an artificial intelligence interactive home shopping telephone ordering method according to an embodiment of the present invention. As illustrated in FIG. 2, in the method for ordering an artificial intelligence interactive home shopping telephone according to an embodiment of the present invention, the legacy server 400 generates the product information of a plurality of products by mapping the home shopping broadcast schedule information of each product. The stored mapping information may be stored (S210), and product information of a product currently being broadcast on home shopping may be extracted from the stored mapping information (S220).

Then, after the product information is extracted, the legacy server 400 compares the schedule information of the extracted product information with the current time (S230), verifies validity (S240), and fails validation (S240-N), It is determined that the product information is not suitable for the current time, and product information can be extracted again (S245).

The interactive voice response server 100, upon receiving the customer's voice input (S250), transmits the voice input to the STT processing module 200, and the STT processing module 200 converts the voice input into text information (S260), it may be transmitted to the artificial intelligence application (300).

The artificial intelligence application 300, in conjunction with the legacy server 400, performs an order procedure for a product based on the converted text information (S270). In the process of ordering the product, it determines whether an event has occurred (S280), when an event occurs (S280-Y), the customer can be connected to the telephone to the telephone agent (S285), if the event does not occur (S280-N), can complete the ordering process of the product (S290) ).

For example, the artificial intelligence application 300 determines that the unstructured content included in the converted text information is not suitable for text analysis using natural language processing, so that the order procedure is not normally performed or the converted text As a result of analyzing sentiment as metadata by analyzing text from unstructured content included in information, and analyzing sentiment for specific phrases and emotions for the entire document, the customer's emotional state is negative emotions such as dissatisfaction, irritability, anger, etc. When an event such as a situation is determined to occur, the customer may be connected to a telephone agent by telephone.

And the artificial intelligence application 300, in conjunction with the legacy server 400, when obtaining the member information for the age of the customer, the event occurrence threshold is set according to the age of the customer, when such an event occurs, the same situation occurs Even so, the higher the age of the customer, the easier the event can occur.

3. Product ordering process

3 is a diagram provided to explain the ordering process of a product by an artificial intelligence interactive home shopping telephone ordering method according to an embodiment of the present invention. As described above with reference to FIG. 2, the artificial intelligence application 300 according to an embodiment of the present invention may perform an order procedure for a product based on the converted text information in conjunction with the legacy server 400.

Specifically, the artificial intelligence application 300 may acquire member information of a customer stored in the legacy server 400 in conjunction with the legacy server 400. In addition, if the customer is not a member, membership information may be acquired through a new member registration, and the new member information may be stored in the legacy server 400.

In addition, as shown in FIG. 3, the artificial intelligence application 300 works with the legacy server 400 to acquire product information of a product currently being broadcast on home shopping, and introduces greetings and product purchase related to the obtained product information. Text information including speech may be transmitted to the TTS processing module 500.

The TTS processing module 500 converts the text including the received greeting and the product purchase related introduction to voice information, delivers it to the interactive voice response server 100, and the interactive voice response server 100 uses voice information. The converted greeting and the introduction words related to product purchase may be delivered to the customer (S310).

And the artificial intelligence application 300, using the STT processing module 200 and the TTS processing module 500, collects the purchase information including the purchase option, quantity and address of the product, and works with the legacy server 400 By doing so, a payment procedure may be performed (S380).

Specifically, the artificial intelligence application 300, the order product confirmation procedure (S320), the order quantity confirmation procedure (S330), the order history confirmation procedure (S340), delivery, such as order specific notice procedure (S350), card, account information For example, a payment method verification procedure (S360) and a delivery address verification procedure (S370) may be sequentially performed.

And the artificial intelligence application 300, at any time during the ordering process, when a product information-related unexpected question is received from the customer, the intention of the natural language-based sentence is to be understood and to provide appropriate answers and feedback (S1), the intention of the sudden question If the grasping fails or the customer repeats the question again, and it is determined that the provision of the proper answer and feedback has failed, the event is determined to occur, and the customer is connected to the telephone agent.

In particular, the artificial intelligence application 300, depending on the product, "What is the expiration date?", "Is card discount possible?" For questions with frequent inquiries, etc., information about expected answers can be extracted and stored in advance to provide appropriate answers and feedback to customer inquiries.

In addition, the artificial intelligence application 300 may process this by executing the order quantity confirmation process again, etc., at any time during the ordering process, when an order quantity change is requested (S2).

And the artificial intelligence application 300, through the legacy server 500, in order to improve the success rate of recognizing address information in the delivery address verification procedure, as well as to record the voice for the address, as well as the text analysis result for the address, When the address recognition fails, it is possible to infer the address information of the upper layer based on the address information of the lower layer, where accuracy of a threshold or higher is secured in the logical hierarchy of addresses.

At this time, the logical hierarchical structure of the address information includes the first tier including the city/city/province, and the city/county/gu/eup/myeon, and the second tier, the second tier, according to the administrative district. /Li is included, the third layer as the lower layer of the second layer, and the barrel/ban/address/lake is included, and may be composed of the fourth layer as the lower layer of the third layer.

That is, the artificial intelligence application 300, if the address of the destination is recognized as "12, Songpa-daero, Songpa-gu, Seoul", if the region matching the address information of'Songpa-gu' is not identified,'Songpa-daero 12' By using the address of the lower layer,'Songpa-gu' can be inferred and recognized as'Songpa-gu'.

Meanwhile, when the payment is completed, the artificial intelligence application 300 may complete the order procedure (S390).

4. Composition of artificial intelligence interactive home shopping phone order server

4 is a diagram provided to explain the configuration of an artificial intelligence interactive home shopping telephone order server according to another embodiment of the present invention. As shown in Figure 4, the artificial intelligence interactive home shopping phone order server 700 according to another embodiment of the present invention includes an interactive voice response processing unit 710, an STT processing unit 720, a processor 730, and a storage unit ( 740) and the TTS processing unit 750.

The interactive voice response processing unit 710 is a means for receiving a voice input of a customer and transmitting voice information in response to the voice of the customer when a call is connected from the customer terminal 10 such as a telephone or a smart phone. The STT processing unit 720 is an information processing means that converts voice input into text information and delivers it to the processor.

The storage unit 740 stores the product information of the product, and the algorithm, program, and information necessary for the interactive voice response processing unit 710, the STT processing unit 720, the processor 730, and the TTS processing unit 750 to operate normally. It is a storage medium that stores them.

The TTS processing unit 750 is an information processing means for converting the written text information into voice information implemented in various languages and voices.

The processor 730 extracts product information of a product currently being broadcast on home shopping among stored product information, provides an appropriate answer and feedback to the customer's questions and requirements based on the converted text information, and performs an order procedure for the product Information processing means.

The processor 730 may access the artificial intelligence server 20 to use artificial intelligence algorithms, analyze text from unstructured content included in the converted text information using a natural language understanding algorithm, and express emotions and documents for specific phrases You can extract sentiment that analyzes sentiment for the whole.

In addition, the processor 730 determines whether an event occurs for a situation in which the order procedure is not normally performed in the process of ordering a product based on text information, so that when an event occurs, the customer is connected to a telephone agent Thus, it is possible to minimize the inconvenience of the customer.

On the other hand, the technical idea of the present invention can be applied to a computer-readable recording medium containing a computer program that performs functions of the apparatus and method according to the present embodiment. Further, the technical idea according to various embodiments of the present invention may be implemented in the form of computer-readable codes recorded on a computer-readable recording medium. The computer-readable recording medium can be any data storage device that can be read by a computer and stores data. Of course, the computer-readable recording medium may be a ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical disk, hard disk drive, and the like. In addition, computer-readable codes or programs stored on a computer-readable recording medium may be transmitted through a network connected between computers.

In addition, although the preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the specific embodiments described above, and the technical field to which the present invention pertains without departing from the gist of the present invention claimed in the claims. In addition, various modifications can be made by those skilled in the art, and these modifications should not be individually understood from the technical idea or prospect of the present invention.

10: customer terminal
20: artificial intelligence server
100: interactive voice response (Interactive Voice Response) server
200: STT (Speech to Text) processing module
300: artificial intelligence application
400: legacy server
500: TTS (Text To Speech) processing module
600: linked server
700: artificial intelligence interactive home shopping phone order server
710: Interactive voice response processing unit
720: STT (Speech to Text) processing unit
730: processor
740: storage unit
750: TTS (Text To Speech) processing unit

Claims (12)

Extracting product information of a product currently being broadcast on home shopping;
Receiving a voice input from a customer;
Converting the voice input into text information; And
Artificial intelligence interactive home shopping phone ordering method comprising; performing a procedure for ordering a product based on the converted text information.
The method according to claim 1,
Step of extracting the product information,
Storing mapping information generated by mapping product information of a plurality of products and home shopping broadcast schedule information of each product; And
Extracting the product information of the product currently being broadcast home shopping among the stored mapping information; artificial intelligence interactive home shopping phone ordering method comprising a.
The method according to claim 2,
Step of extracting the product information,
After the product information is extracted, comparing the schedule information of the extracted product information with the current time to verify validity; And
When the validation fails, determining that the product information is not suitable for the current time, and extracting the product information again; artificial intelligence interactive home shopping phone ordering method further comprising.
The method according to claim 1,
Steps to perform the order process of the product,
Artificial intelligence interactive home shopping phone ordering method characterized by determining whether an event has occurred in the process of performing an order process based on the converted text information and, when an event occurs, a customer is connected to a telephone agent.
The method according to claim 4,
The event,
An artificial intelligence interactive home shopping telephone ordering method characterized in that the unstructured content included in the converted text information is judged to be unsuitable for text analysis using natural language processing and the order procedure is not normally performed.
The method according to claim 4,
The event,
As a result of analyzing sentiment as metadata by analyzing sentiment as metadata through text analysis from unstructured content included in the converted text information, the customer's emotional state is dissatisfied, irritable, or angry Artificial intelligence interactive home shopping phone ordering method characterized in that the situation is judged as one.
The method according to claim 4,
The event,
Acquiring member information about the age of the customer, even if the same situation occurs, the higher the age of the customer, the easier the event occurs, so that the event generation threshold is set according to the age of the customer, according to the AI conversation How to order an older home shopping phone.
The method according to claim 1,
The steps to perform the ordering process are:
Obtaining customer's member information;
Converting greetings and introductions related to product purchase into voice information;
Collecting purchase information including purchase options, quantities and addresses of products; And
Performing a payment procedure; Artificial intelligence interactive home shopping phone ordering method comprising a.
The method according to claim 8,
The step of collecting purchase information is,
As a result of the text analysis on the address, when the recognition of the address fails, the artificial intelligence interactive home shopping is characterized by inferring the address information of the upper layer based on the address information of the lower layer having an accuracy above a threshold in the logical hierarchical structure of the address. How to order by phone.
The method according to claim 9,
The logical hierarchy of address information is,
Depending on the administrative area, the first tier including the city/city/province, city/gun/gu/eup/myeon is included, and the second tier, dong/ri, which is the lower tier of the first tier, and the lower tier of the second tier A method of ordering an artificial intelligence interactive home shopping phone, characterized in that it comprises a third layer, which is a layer, and a traffic/ban/address/lake, and a fourth layer, which is a lower layer of the third layer.
A legacy server that extracts product information of products currently being broadcast on home shopping;
An interactive voice response server that receives a customer's voice input;
A speech to text (STT) processing module that converts speech input into text information; And
Artificial intelligence interactive home shopping telephone ordering system, including; an artificial intelligence application that performs an order procedure for a product based on the converted text information in conjunction with a legacy server.
A storage unit in which product information of products is stored;
An interactive voice response processing unit that receives a customer's voice input;
A speech to text (STT) processing unit that converts voice input into text information; And
Artificial intelligence interactive home shopping telephone order server, including; a processor for extracting product information of a product currently being broadcast on home shopping among stored product information and performing an order procedure for the product based on the converted text information.

Images