JP6709132B2 - Voice order reception system - Google Patents

Description

The present invention relates to a technology that supports acceptance of orders and reservations, and more particularly, to technology that is effective when applied to a voice order acceptance system that supports acceptance of orders and reservations by voice.

For example, when a securities company receives an order to buy or sell a stock from an individual customer, online order reception via a website has become widespread in recent years, but order reception by telephone is mainly performed by elderly customers and the like. Has also been done.

In a securities company, when a customer receives an order to buy or sell a stock by telephone, the operator at the window for accepting the order inputs the order details on an order screen displayed on a terminal used by the operator. Furthermore, the input content is read aloud to confirm with the customer. It takes a considerable amount of time to perform such processing, and this is one of the causes of the long response time for the customer as a whole. In addition, incorrect input leads to erroneous ordering, which requires careful processing, which imposes a mental burden on the operator. Regarding this point, for example, although the input quality can be improved by checking the input value, the current situation is that no further technical support can be provided.

As a technique related to ordering by voice, for example, Japanese Unexamined Patent Application Publication No. 2007-4282 (Patent Document 1) captures a utterance content of a customer from a voice signal received from a voice input/output device provided in a seat of a restaurant. It is described that by determining whether or not the necessary information is available as the order information, the customer can place an order directly without going through the clerk.

On the other hand, as a technique for grasping the structure of the conversation from the voice data of the conversation, for example, Japanese Patent Laid-Open No. 2011-134291 (Patent Document 2) discloses a flow in a flow definition set in advance for each statement in the discourse data. It describes that when the expression matches, the corresponding flow is set and output as discourse semantics. Further, it is also described that the proper expression can be analyzed by a rule base based on a preset proper expression analysis rule, and the proper expression information in each statement can be output as the discourse semantics.

JP, 2007-4282, A JP, 2011-134291, A

For example, by using the conventional technique described in Patent Document 1, it is possible to realize a configuration in which a customer receives an order by voice such as a telephone call. However, in the conventional technique, one unit of the conversation ends when the information necessary to complete the order is gathered, and the order is processed in the server by using this as a trigger. Therefore, the order cannot be fulfilled in the state where the necessary information is insufficient. Further, when ordering a plurality of products or the like, it is necessary to utter all the information necessary to complete an order for each product or the like and complete one unit of the conversation. Therefore, in order to establish a plurality of orders related to a plurality of products or the like with similar or similar contents, the customer may have to repeatedly issue the same wording.

Therefore, an object of the present invention is to enable an operator to efficiently receive a plurality of orders relating to a plurality of products and the like when an operator receives an order by voice such as a telephone call from a customer, and is necessary for establishment of an order. An object of the present invention is to provide a voice order reception system that supports supplementation when information is insufficient.

The above and other objects and novel features of the present invention will be apparent from the description of this specification and the accompanying drawings.

The following is a brief description of the outline of the typical inventions among the inventions disclosed in the present application.

A voice order reception system according to a representative embodiment of the present invention is a voice order reception system that supports an operator who receives voice orders or reservations from a customer to input order information into the order system. The processing includes a morpheme analysis unit that divides each statement in the conversation data in which the content of the conversation regarding the order or reservation between the customer and the operator into a text into morphemes and determines a part of speech by processing.

Furthermore, a regular expression for identifying a specific expression in each statement in the conversation data that has been subjected to morphological analysis, and an order information dictionary composed of a correspondence between the specific expression corresponding to a part matching the regular expression, From each of the statements in the conversation data subjected to morphological analysis, the specific expressions corresponding to the respective elements necessary for establishing the order or reservation are extracted based on the order information dictionary, respectively, and as the order information, And an order information extraction unit to be displayed on the first screen of the information processing terminal used by the operator.

The effects obtained by the typical ones of the inventions disclosed in the present application will be briefly described as follows.

That is, according to the representative embodiment of the present invention, when the operator receives an order by voice such as a telephone call from a customer, it is possible to efficiently receive a plurality of orders for a plurality of products or the like. At the same time, it becomes possible to support the replenishment when the information required to complete the order is insufficient.

It is a figure showing the outline about the example of composition of the voice order acceptance system which is one embodiment of the present invention. It is the figure which showed the outline about the example which extracts order information from the conversation in one embodiment of this invention. It is the figure which showed the outline about the example which inputs and cooperates with order information in one embodiment of this invention. It is the flowchart which showed the outline about the example of the flow of the process which extracts order information from the conversation in one embodiment of this invention, and places an order. It is the figure which showed the outline about the example of the various patterns which extract order information from the conversation in one embodiment of this invention. It is the figure which showed the outline about the other example of the various patterns which extract order information from the conversation in one embodiment of this invention. It is the figure which showed the outline about the other example of the various patterns which extract order information from the conversation in one embodiment of this invention. It is the figure which showed the outline about the other example of the various patterns which extract order information from the conversation in one embodiment of this invention. It is the figure which showed the outline about the other example of the various patterns which extract order information from the conversation in one embodiment of this invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In all the drawings for explaining the embodiments, the same parts are denoted by the same reference symbols in principle and the repeated description thereof will be omitted. On the other hand, parts described with reference numerals in a certain drawing may be referred to with the same reference numeral, although not shown again, in the description of other drawings.

<System configuration>
FIG. 1 is a diagram showing an outline of a configuration example of a voice order receiving system according to an embodiment of the present invention. The voice order receiving system 30 according to the present embodiment is, for example, an information processing system used in the call center 1 of a securities company that receives a telephone order for stock trading from a customer 2.

The voice order receiving system 30 is composed of, for example, a server system including a server device and a virtual server constructed on a cloud computing service. The information processing terminal is not limited to the server system, and may be configured by an information processing terminal such as a PC (Personal Computer). Then, a CPU (Central Processing Unit) (not shown) executes middleware such as an OS (Operating System) and a DBMS (DataBase Management System) expanded on a memory (not shown), and software that operates on the middleware, which will be described later. To realize the function of each part.

The voice order reception system 30 includes, for example, a voice recognition engine 31, a morpheme analysis unit 32, an order information extraction unit 33, and the like implemented by software. It also has a data store of the order information dictionary 35 implemented by a database, a file table, or the like.

The voice recognition engine 31 receives a telephone call from the customer 2 and processes the conversation via the PBX (Private Branch eXchange)/CTI (Computer Telephony Integration) 10 for a stock order between the customer 2 and the operator 20. It has a function of acquiring the voice data of, converting it into text data in real time, and outputting it as the conversation data 34. An existing/known one can be appropriately used as the voice recognition engine 31. It goes without saying that the higher the recognition rate of the voice recognition engine 31, the more accurate the order reception process can be performed.

The output (or output) conversation data 34 is subjected to processing such as data layout conversion such as preprocessing in the discourse structure analysis system described in Patent Document 2 described above, if necessary. May be. For example, processing is performed by grouping consecutive statements for each speaker as a block, assigning a sequence number (ID) indicating the order to each block and each statement, and extracting/setting the speaker data of each block. To do. The conversation data 34 after the processing is represented as a data structure using, for example, XML (eXtended Markup Language). Further, in addition to this, analysis of the conversation structure (analysis of the semantic content of each statement) such as the flow analysis process in the discourse structure analysis system described in Patent Document 2 may be performed.

The morpheme analysis unit 32, in the same manner as the morpheme analysis process in the discourse structure analysis system described in Patent Document 2, arranges each statement in the conversation data 34 in a column of morphemes (the smallest unit having a meaning in the language). It has a function of performing a so-called morphological analysis that divides and determines a part of speech and outputs the analysis result. The analysis result holds, for example, the ID of each statement in the conversation data 34 and the information of the morpheme string corresponding thereto, and is represented as a data structure using XML or the like. An existing/known morpheme analysis engine (for example, a free one such as MeCab (Wabu Kabu) or ChaSen (Chasen) or another commercial system) can be used as the morpheme analysis unit 32.

The order information extraction unit 33 receives the conversation data 34 (or the morpheme sequence information and the conversation data 34) to which the morpheme sequence information 32 is attached by the morpheme analysis unit 32 as input, and also refers to the order information dictionary 35 to obtain stock information in real time. It has a function of extracting a word corresponding to an element necessary for the establishment of the sales order and outputting it as the order information 36.

To extract each word of the order information 36 from the information of the morpheme string and the conversation data 34, for example, the same processing as the proper expression analysis processing in the discourse structure analysis system described in Patent Document 2 can be performed. Specifically, for example, based on a proper expression analysis rule registered in advance in the order information dictionary 35, the conversation data 34 is analyzed by a rule base using a regular expression, and necessary for establishing a trading order. An elemental unique expression is extracted as order information 36. Here, the proper expression analysis rule is a list of correspondences between the regular expressions for specifying the proper expressions in each statement in the conversation data 34 and the proper expressions corresponding to the parts matching the regular expressions.

The unique expression extracted here is a character string including a proper noun, a numerical value, and the like. In this embodiment, for example, a company name (including an abbreviation), a stock code, and other terms used in ordering (such as “limit price” and “market”) are factors that can be necessary for establishing a stock sales order. Etc.) etc. shall be included.

The order information 36 extracted from the conversation data 34 is displayed on the screen of the operator terminal 21 by the order information extraction unit 33, as will be described later, and receives an instruction from the operator 20 to the order system 3 as necessary. May be set as a parameter when inputting the order. Alternatively, the order information extraction unit 33 may directly set the order as a parameter of the order input to the order system 3 without automatically outputting the order to the operator 20 via the operator terminal 21, and automatically generate the order.

<Extracting order information from conversation>
FIG. 2 is a diagram showing an outline of an example of extracting the order information 36 from the conversation according to the present embodiment. The upper diagram of FIG. 2 shows an example of a conversation between the customer 2 and the operator 20 regarding a stock order by telephone. In the middle diagram, the conversation shown in the example in the upper diagram is voice-recognized by the voice recognition engine 31 of the voice order receiving system 30 and converted into text, and the statements for each speaker are summarized as a block, and then the conversation is performed. The example of the result of having analyzed the structure (flow) is shown typically.

Then, the lower diagram shows an example of the result of extracting the unique expressions that can be the order information 36 by the order information extracting unit 33 from the conversation structure shown in the example of the middle diagram. In the example in the figure, from the utterance of the customer 2, the "stock", "order type (buying or selling)", "quantity", "unit price (limit price or market)" required to complete a stock buy/sell order It indicates that the information (proper expression enclosed by a solid line frame) related to each element is extracted.

FIG. 3 is a diagram showing an outline of an example of inputting and linking the order information 36 in the order system 3 according to the present embodiment. The upper diagram is an example of a screen displaying the extraction result of the order information 36 displayed on the operator terminal 21 by the order information extracting unit 33 of the voice order receiving system 30. Here, the contents of the order information 36 extracted in the example of FIG. 2 are displayed for each of the elements (“trading (order type)”, “brand”, “quantity”, and “unit price”) necessary for establishing a trading order for stocks. It has been shown that. A button is associated with each element on the screen. For example, when the operator 20 presses the "buy order screen display" button by clicking or tapping, the ordering system 3 as shown in the example in the lower diagram is provided by executing the corresponding script included in the screen. Display the stock purchase order input screen.

After that, the operator 20 returns to the order information extraction screen shown in the upper diagram on the operator terminal 21 and clicks the "transfer" button corresponding to each of the other elements ("brand", "quantity", "unit price") as necessary. Press down. When the operator 20 presses the “transfer” button, the value or content of each element is displayed in the corresponding input field of the buy order input screen provided by the ordering system 3 shown in the lower diagram by, for example, executing the corresponding script. Transfer and copy. At this time, if necessary, for example, processing such as conversion from brand name to brand code may be automatically performed as illustrated. Further, the operator 20 may correct the content of the extracted order information 36 as necessary and then press the “transfer” button.

In this way, the order information 36 extracted as the value of each element necessary for the establishment of an order can be transferred to the order input screen of the order system 3 under the judgment of the operator 20. As a result, for example, if there is a mistake in voice recognition, a mistake in extracting a proper expression, a mistake in the customer 2 or the operator 20, or the like, the inappropriate content of the order information 36 is directly used for the actual order to the order system 3. It is possible to prevent being caught.

FIG. 4 is a flowchart showing an outline of an example of a flow of processing for extracting order information 36 from a conversation and placing an order in the present embodiment. In the voice order receiving system 30, when voice data of a telephone conversation relating to an order between the customer 2 and the operator 20 is acquired from the PBX/CTI 10, first, this is voice-recognized by the voice recognition engine 31 in real time and the text is recorded. It is converted and output as conversation data 34 (S01). At this time, as described above, if necessary, processing such as data layout conversion such as preprocessing in the discourse structure analysis system described in Patent Document 2 and analysis of conversation structure such as flow analysis processing are performed. Processing may be performed.

After that, the morpheme analysis unit 32 divides each statement in the conversation data 34 into a morpheme sequence in real time and performs morpheme analysis to determine the part of speech (S02). This processing may be performed intermittently, or may be performed in units of one sentence in the conversation data 34. After that, the order information extracting unit 33 extracts a proper expression from the analyzed conversation data 34 and outputs it as order information 36 (S03). At this time, it is determined from the conversation structure of each statement in the conversation data 34 whether or not the conversation breaks (S04). A case where the conversation is complete also when all of the elements necessary for the establishment of an order are prepared as the order information 36.

If the conversation does not correspond to a break, that is, if the conversation related to the order continues (S04: No), the process returns to step S01 and the series of processes is continued. As will be described later, multiple orders may occur in one conversation (statement). In this case, it is necessary to perform the subsequent processing when all the necessary elements for each order are completed. Because there is.

When the conversation corresponds to a break, that is, when the conversation regarding the order ends (S04: Yes), the order information extraction unit 33 displays the order information 36 extracted in step S03 in the upper part of FIG. The order information extraction screen as shown in is displayed on the operator terminal 21 (S05). The operator 20 operates the order input screen of the order system 3 by inputting the order information 36 corresponding to each element necessary to complete the trading order by pressing a button or the like in the example in the upper part of FIG. 3 as necessary. It is possible to make a buy/sell order for the ordering system 3 by linking with the (S06).

<Example of order information extraction>
5 to 9 are diagrams showing an outline of examples of various patterns for extracting the order information 36 from the conversation in the present embodiment. The upper diagram of FIG. 5 shows an example of a conversation regarding a stock trading order between the customer 2 and the operator 20. Here, among the elements necessary for fulfilling a sales order, the items “sales (order type)”, “quantity”, and “unit price” can be extracted from the utterance of the customer 2. It shows that it is possible (“I want to buy”, “200 shares”, “3000 yen”, and “limit price” surrounded by a solid line frame in the figure).

On the other hand, with respect to the "brand", the customer 2 only reports "Nomura" (the part surrounded by the broken line frame in the figure), and the specific brand cannot be uniquely determined only by this information. For example, in addition to "Nomura Securities", other companies such as "Nomura Research Institute" and "Nomura Real Estate", group companies, affiliated companies, etc. may be included in the stocks labeled "Nomura." In this way, when it is not possible to determine the specific content as the order information 36 for each element required to fulfill the order (when a plurality of candidates hit), in the present embodiment, the lower diagram of FIG. As shown in, for example, with respect to the “brand” element, the character string “Nomura” of the proper expression extracted from the conversation data 34 is highlighted or highlighted so as to be emphasized. At this time, the corresponding "transfer" button may be disabled.

As a result, the operator 20 confirms that some candidates are relevant to each element necessary to complete the buy/sell order, and are in an ambiguous and undecidable state (that is, a state in which the confirmed order information 36 is insufficient). Can be recognized. Recognizing this, the operator 20 can ask the customer 2 additional questions to uniquely determine the content of the ambiguous element.

The upper diagram of FIG. 6 shows an example of the conversation when the operator 20 further issues an additional question following the conversation in the upper diagram of FIG. Here, the operator 20 issues an additional question (confirmation) in response to an utterance of an order in which the content of the “brand” by the customer 2 is ambiguous, and among them, the unique expression relating to the “brand” (in the figure, “ It is shown that the confirmation question including the “Nomura Research Institute” (“...?”) is being uttered. It also indicates that the customer 2 has accepted the confirmation question (“Yes.”). In this case, for example, as shown in the “Order 1” portion of the lower diagram of FIG. 6, the contents related to the “brand” element in the order information 36 of each element displayed on the order information extraction screen are The order information 36 ("Nomura Research Institute") extracted from the utterance of the confirmation question of the operator 20 is overwritten.

Further, the upper diagram of FIG. 6 shows that the customer 2 is making a utterance (“later...”) regarding an additional order. In this case, it is not determined here that there is a break in the conversation, but it is determined that there is a break in the conversation when the conversation related to the additional order ends. Then, as shown in the lower diagram of FIG. 6, an order information extraction screen displaying the order information 36 extracted for the two orders (“order 1” and “order 2”) is displayed. In the example of FIG. 6, two orders are displayed side by side in the vertical direction, but the display format is not particularly limited, and a plurality of orders may be displayed side by side in the horizontal direction, or a tab screen may be used to display the orders. The order displayed by 20 may be switched.

As for the additional order (“Order 2”), as shown in the utterance of Customer 2 in the upper diagram, where it is the “buy” order of the “limit price”, the “stock” (“○△□”) and the “order” Although the "quantity" ("100 shares") is specified, the "unit price" itself in the "limit price" is not specified. In this case, as shown in "Order 2" in the lower diagram, the fact that the order information 36 is insufficient for the "unit price" amount portion is highlighted by highlighting or the like. Recognizing this, the operator 20 can ask further questions to the customer 2 in order to obtain the missing information.

FIG. 7 shows an example of the conversation when the operator 20 further issues an additional question for an additional order (“Order 2”) following the conversation in the upper part of FIG. Here, in response to the utterance of the "limit price" order in which the customer 2 does not specify the "unit price" amount, an additional question ("... how much is it? ?”) is being emitted. Then, in response to the inquiry about the market price from the customer 2 (“How much is it now?”), the operator 20 answers the market price (“...2315 yen”), and in response to this, the customer 2 receives the “limit price”. This indicates that the conversation is being performed in the manner of designating the "unit price" ("... 2200 yen.") and the operator 20 confirming this ("... 2200 yen limit.").

In the present embodiment, such a conversation structure is grasped by, for example, a flow analysis in the discourse structure analysis system described in Patent Document 2, and the order information 36 is appropriately extracted. That is, in the above case, the amount of the "unit price" in the "limit price" is not "2315 yen" which the operator 20 answered as the market price, but "2200 yen" confirmed by the operator 20 by the customer 2 as the "unit price". Is extracted as order information 36. In the "Order 2" portion of the order information extraction screen shown in FIG. 8, "2200 yen" extracted as the order information 36 is added and reflected as the "unit price".

Further, returning to FIG. 7, it is further shown that the customer 2 is making an utterance related to the change of the content of the previous order (that is, “order 1”) (“after all the previous one...”). There is. In this case as well, the structure of the conversation is grasped by, for example, the flow analysis in the discourse structure analysis system described in Patent Document 2, which order is appropriately specified and the order information 36 is extracted. That is, in the above case, it is determined that "Order 1" is the target, based on the utterance of "the one before the customer" of the customer 2 and the determination that the current conversation target is "Order 2". Then, with respect to the target "order 1", the customer 2 makes an utterance that the "limit" is changed to the "market".

On the other hand, the operator 20 has issued an additional question for confirming the content of the order change, in which the proper expression (“Nomura Research Institute” in the figure) relating to “stock” and “unit price” The confirmation question is uttered ("Changed to ...?") that includes the proper expression ("Case" in the figure). It also indicates that the customer 2 has accepted the confirmation question (“Yes, that's right.”). In this case, for example, as shown in the “Order 1” portion of FIG. 8, the operator 20 confirms the content related to the “unit price” element in the order information 36 of each element displayed on the order information extraction screen. The order information 36 (“market”) extracted from the utterance of the question is overwritten.

The upper diagram of FIG. 9 shows an example of a conversation when the customer 2 collectively makes a plurality of sales orders. Here, the customer 2 has made a “buying” order in “market” for a plurality of “brands”, and among the respective elements required for order fulfillment, “brand” and “quantity” are respectively Are individually specified for the orders (“Nomura Research Institute” “200 shares” and “○△□” “100 shares” in the figure). On the other hand, regarding “buying/selling (order type)” and “unit price” among the respective elements, the fact that it is a “buying” order in “market” is uttered only once. Therefore, the fact that the order is a "buy" order on the "market" is interpreted as being related to each order.

Therefore, in the present embodiment, as shown in the lower diagram of FIG. 9, for each order (“Order 1” and “Order 2” in the figure), both are “buy” orders in “market”. , Complements/supplements the contents of the order information 36. As a result, it is possible to greatly reduce the work load on the operator 20 and the customer 2 relating to the stock trading order.

In the example of FIG. 9, the fact that it is a “buy” order in “market” is complemented/supplemented by copying it to both “order 1” and “order 2”. It is not limited to copying the same contents in this way. Based on the content of the order information 36 extracted in each order, the type of the target element, and the like, the appropriate content may be estimated and converted, and then supplemented/supplemented. If the appropriate content cannot be guessed, it is assumed that the order information 36 is insufficient, and the fact that the target element is insufficient is highlighted by highlighting or the like as in the example of the lower diagram of FIG. May be.

As described above, according to the voice order receiving system 30 according to the embodiment of the present invention, voice data of a conversation with the operator 20 regarding a telephone order from the customer 2 is converted into a text in real time, and the order is placed. It is possible to extract, as real-time order information 36, a proper expression that is the content of an element necessary for establishment. The operator 20 can use the extracted order information 36 as it is or by appropriately modifying it, when inputting an order to the order system 3.

In addition, when multiple orders for multiple products are made with one utterance (statement), when common utterances for multiple orders are uttered only once, this is referred to as another order. Can be replenished and supplemented as the order information 36. In addition, when the contents of the elements required for order fulfillment cannot be complemented/supplemented or cannot be uniquely determined, the fact is displayed so that the operator 20 can recognize the fact, and the operator 20 adds It is possible to easily ask questions. As a result, the physical and psychological burden on the operator 20 for receiving orders can be greatly reduced.

Although the invention made by the present inventor has been specifically described based on the embodiments, the present invention is not limited to the above embodiments, and various modifications can be made without departing from the scope of the invention. Needless to say. For example, the above embodiments have been described in detail for the purpose of explaining the present invention in an easy-to-understand manner, and are not necessarily limited to those having all the configurations described. Further, it is possible to add/delete/replace other configurations with respect to a part of the configurations of the above-described embodiments.

For example, in the above-described embodiment, the order for stocks in a securities company or the like is targeted, but the operator 20 may accept a voice order or reservation from the customer 2 by telephone or the like. For example, orders/reservations for products and services other than stocks may be targeted.

Further, in the above embodiment, the operator 20 accepts a telephone order from the customer 2 and extracts the order information 36. When there is a shortage, the operator 20 calls the customer 2 with an additional question. However, the method is not limited to such a response method. For example, it is combined with an automatic dialogue system (which may be publicly known) that analyzes the utterance content of the customer 2 in real time and can automatically select and generate the response content such as an appropriate question or answer. It is also possible. As a result, it is possible to automatically generate an order while supplementing and replenishing the order information 36 by automatically asking additional questions without the help of the operator 20.

Each of the configurations, functions, processing units, processing means, and the like in the above-described embodiments may be implemented in hardware by designing a part or all of them in an integrated circuit, for example. Further, each of the above-described configurations, functions, and the like may be realized by software by a processor interpreting and executing a program that realizes each function. Information such as a program, a table, and a file that realizes each function can be stored in a recording device such as a memory, a hard disk, an SSD (Solid State Drive), or a recording medium such as an IC card, an SD card, and a DVD.

Further, in each of the above drawings, the control lines and information lines are shown to be necessary for explanation, and not all the control lines and information lines for mounting are necessarily shown. In practice, it may be considered that almost all configurations are connected to each other.

INDUSTRIAL APPLICABILITY The present invention can be used for a voice order reception system that supports reception of voice orders and reservations.

1...Call center, 2...Customers, 3...Order system,
10...PBX/CTI,
20...Operator, 21...Operator terminal,
30... Voice order reception system, 31... Voice recognition engine, 32... Morphological analysis unit, 33... Order information extraction unit, 34... Conversation data, 35... Order information dictionary, 36... Order information

Claims (8)

A voice order reception system comprising a computer system that supports the input of order information to an order system by an operator who receives voice orders or reservations from customers,
A morpheme analysis unit that determines a part of speech by dividing each statement into morphemes for each statement in the conversation data in which the content of the order or reservation conversation between the customer and the operator is converted to text by voice recognition processing,
Order information consisting of a correspondence between a regular expression for specifying a specific expression in each statement in the conversation data that has been morphologically analyzed by the morphological analysis unit and the specific expression corresponding to a portion matching the regular expression. Dictionary and
From the statements in the conversation data morphologically analyzed by the morpheme analysis unit , based on the order information dictionary, the unique expression corresponding to each element of a plurality of elements necessary for establishing the order or reservation extracted respectively, have a, and order information extracting unit to be displayed on the first screen of the information processing terminal that the operator uses as order information,
The order system displays the order information element designated by the operator among the order information elements displayed on the first screen for inputting the order or reservation. Enter to transfer to the second screen,
The order information extraction unit, when extracting the order information, determines the break of the conversation on the time series of the conversation, and for each conversation portion corresponding to the break of the conversation, the order information of the first screen is displayed. The elements of the order information, which is the display content, are automatically determined and the display of the first screen is updated, and the display content of the first screen is the display of the missing or unknown elements among the plurality of elements. And may be updated to automatically overwrite or supplement the missing or unknown element with information extracted in response to the subsequent conversation portion.
Voice order reception system. In the voice order receiving system according to claim 1 ,
Was converted so as to correspond to the items of the screen element said order based on the information dictionary of the second specified the order information from the operator of the elements of the order information displayed on the first screen to enter to transfer to the second screen above, the voice order receiving system. In the voice order receiving system according to claim 1,
The order information extraction unit, on the basis of the extracted order information, the conversation when the data contains the order or reservation multiple number, order information, which is only spoken once common to a plurality of the order or reserved A voice order reception system that complements elements by reflecting them on all of a plurality of orders or reservations. In the voice order receiving system according to claim 1,
When the extracted order information is insufficient for a plurality of elements necessary for establishing the order or reservation, the order information extraction unit relates to the insufficient target element on the first screen. with emphasis display, the transfer for the elements of the object to inoperable state operation item for specifying by the user, updated results, lack the elements based on the extracted from subsequent conversation part A voice order receiving system that returns the operation item to an operable state when the state of being canceled is canceled . In the voice order receiving system according to claim 1,
The order information extracting unit, when extracted the order information, not uniquely determined elements of the plurality of elements required for establishment of the order or reserved, in the first screen, the uniquely with highlight display according to the determined not to target element, the transfer for the elements of the object to inoperable state operation item for specifying by the user, based on the extracted from subsequent conversation part A voice order receiving system which returns the operation item to an operable state when the element that is not uniquely determined is resolved as a result of updating the element . In the voice order receiving system according to claim 1,
The conversation the order information extraction unit, including with respect to the first screen for displaying the order information extracted based on the first conversation portion of the conversation, the answer with additional questions and the customer due to subsequent said operator The voice order receiving system which automatically updates the display content so as to overwrite or complement the corresponding element of the order information in the first screen by the unique expression extracted based on the second conversation part of . In the voice order receiving system according to claim 1,
When the conversation data includes a plurality of the orders or reservations based on the extracted order information, the order information extraction unit displays the order for each of the orders or reservations in the first screen. Alternatively, a voice order reception system that separately displays information on multiple elements required for reservation. A voice order reception system including a computer system that supports input of order information to an order system by an automatic dialogue system that receives voice orders or reservations from a customer,
A morpheme analysis unit that divides a morpheme into each morpheme for each statement in the conversation data in which the content of the order or reservation conversation between the customer and the automatic dialogue system is converted into text by voice recognition processing, and
Order information consisting of a correspondence between a regular expression for identifying a specific expression in each statement in the conversation data that has been morphologically analyzed by the morphological analysis unit, and the specific expression corresponding to a portion matching the regular expression. Dictionary and
From each of the statements in the conversation data subjected to morpheme analysis by the morpheme analysis unit, based on the order information dictionary, the unique expression corresponding to each element of the plurality of elements necessary for establishing the order or reservation And an order information extraction unit that extracts each and configures the first information as order information,
Transferring the element of the order information designated by the automatic dialogue system among the elements of the order information in the first information to the second information for inputting the order or reservation into the order system. Input,
The order information extraction unit, when extracting the order information, determines the break of the conversation on the time series of the conversation, and for each conversation part corresponding to the break of the conversation, the order information is extracted. The element of the order information, which is the content, is automatically determined to update the content of the first information, and the content of the first information may include a missing or unknown element among the plurality of elements. And updating to automatically overwrite or supplement the missing or unknown element with information extracted in response to the subsequent conversation portion,
Voice order reception system.