JP2014186061A - Information processing device and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To discriminate whether an important word issued in a telephone conversation is recorded in an appropriate place in a text document recording contents of the telephone conversation.SOLUTION: An association execution part (4d) associates each of a plurality of sentences included in telephone conversation summary data representing a text document with any spoken voice among a plurality of spoken voices represented by telephone conversation voice data. Then, an important word existence/absence determination part (4h) determines whether a sentence including a character string of the important word exists between a sentence associated with the preceding spoken voice of an important spoken voice being a spoken voice including the voice of the important word among the plurality of spoken voices and a sentence associated with a spoken voice after the important spoken voice.

Description

  The present invention relates to an information processing apparatus and a program.

  In Patent Literature 1 below, an input item that should be input by the operator among a plurality of input items is identified from keywords included in the call voice of the operator, and data is input to the specified input item. It is described that it is determined whether or not.

JP 2012-32562 A

  An object of the present invention is to determine whether or not an important word issued in a call is described in an appropriate place in a text document in which the contents of the call are recorded.

  In order to solve the above problem, the information processing apparatus according to claim 1, wherein the call is indicated by a result of morphological analysis processing for each of a plurality of unit texts included in a text document in which the content of the call is recorded and voice data. Based on the result of the speech recognition processing for each of the plurality of uttered voices, the association means for associating each unit text with one of the uttered voices, and a predetermined important word of the plurality of uttered voices A first text that is a unit text associated with an utterance voice before the important utterance voice, which is an utterance voice including the voice of the second voice, and a second text that is a unit text associated with the utterance voice after the important utterance voice; Determining means for determining whether or not there is a unit text including the character string of the important word.

  The information processing apparatus according to claim 2 is the information processing apparatus according to claim 1, wherein a unit text including the character string of the important word is present between the first text and the second text. When it is determined not to do so, the fact is output.

  The information processing apparatus according to claim 3 further includes a display control unit that causes the information processing apparatus according to claim 2 to display the text document on a display unit, wherein the display control unit includes the first control unit. When there is no unit text including a character string of the important word between the text and the second text, information indicating the important word is displayed on the display unit together with the text document. .

  Further, an information processing apparatus according to claim 4 is the information processing apparatus according to claim 3, wherein a character string of the important word and a predetermined relationship are set between the first text and the second text. Means for determining whether or not there is a unit text including an alternative character string that is a character string having the character string, and the display control means includes the alternative character between the first text and the second text. When there is a unit text including a column, information indicating the important word and the alternative character string is displayed on the display unit together with the text document.

  Further, in the information processing device according to claim 5, when the information processing device according to claim 4 includes a unit text including the substitute character string between the first text and the second text, Means for updating the text document and correcting the substitute character string included in a unit text existing between the first text and the second text into a character string of the important word; It is characterized by.

  An information processing apparatus according to claim 6 is an information processing apparatus according to any one of claims 1 to 5; independent word extraction means for extracting an independent word included in each unit text by morphological analysis processing; And an independent word specifying means for specifying an independent word extracted by the independent word extraction means, which is uttered in each utterance voice by the speech recognition processing, and the associating means each unit text, It is characterized by associating with the uttered voice in which the same independent word contained in the unit text is uttered.

  An information processing apparatus according to a seventh aspect further includes a voice acquisition unit that acquires the voice data indicating the plurality of uttered voices in the information processing apparatus according to any one of the first to sixth aspects. It is characterized by that.

  In order to solve the above-described problem, the program according to claim 8 is configured to generate a call in the call indicated by the result of the morphological analysis processing for each of the plurality of unit texts included in the text document in which the content of the call is recorded and the voice data. An association means for associating each unit text with one of the uttered voices based on the result of the voice recognition processing for each of the plurality of uttered voices, including a voice of a predetermined important word among the plurality of uttered voices Between the first text that is the unit text associated with the speech before the important speech that is the speech and the second text that is the unit text associated with the speech after the important speech, A computer is caused to function as a determination unit that determines whether or not a unit text including a character string of the important word exists.

  According to the first, seventh, and eighth aspects of the present invention, it is possible to determine whether or not an important word issued in a call is described in an appropriate place in a text document that records the contents of the call.

  According to invention of Claim 2, it can alert | report that the important word is not described in the appropriate place in a text document.

  According to the third aspect of the present invention, it can be notified in more detail that the important word is not described in an appropriate place in the text document, as compared with the case where this configuration is not provided.

  According to the invention of claim 4, for example, it is possible to notify that the important word is not correctly described although the important word is described in an appropriate place in the text document.

  According to the invention of claim 5, for example, it is possible to correct an error of an important word described in an appropriate place.

  According to the invention of claim 6, the location in the text document where the important word is to be recorded can be specified more accurately than in the case where the configuration is not provided.

It is a figure which shows an example of the hardware constitutions of information processing apparatus. It is a functional block diagram which shows the function group implement | achieved with information processing apparatus. It is a figure which shows an example of audio | voice data. It is a figure which shows an example of some speech sounds. It is a figure which shows an example of the content of a text document. It is a figure which shows an example of an important list. It is a figure which shows the memory content of a keyword memory | storage part. It is a figure which shows an example of a word lattice. It is a figure which shows the memory content of a word spotting result memory | storage part. It is a flowchart which shows the process performed with information processing apparatus. It is a conceptual diagram which shows a mode that an object text is linked | related with speech sound. It is a flowchart which shows the process performed with information processing apparatus. It is a figure which shows the image displayed on a display part. It is a flowchart which shows the process performed with information processing apparatus. It is a figure which shows the image displayed on a display part.

  Hereinafter, examples of embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a diagram illustrating a hardware configuration of an information processing apparatus 2 according to an embodiment of the present invention. The information processing apparatus 2 is realized as a computer including a control unit 4, a main memory 6, a hard disk 8, a display unit 10, an operation input unit 12, and the like. In the case of this embodiment, the information processing apparatus 2 is used by a manufacturer.

  The control unit 4 is a microprocessor, and executes various types of information processing according to programs stored in the main memory 6. The main memory 6 is realized by a ROM and a RAM, and stores information necessary for various types of information processing in addition to the above programs. Here, the program is read from a computer-readable information storage medium (for example, DVD (registered trademark) -ROM) and stored in the main memory 6. The program may be downloaded via a network and stored in the main memory 6.

  The hard disk 8 stores various information. Information stored in the hard disk 8 will be described later. The display unit 10 is a display such as a liquid crystal display, and displays information according to a command from the control unit 4.

  The operation input unit 12 is a mouse, a keyboard, or the like, and passes a signal indicating the operation content executed by the administrator of the information processing apparatus 2 to the control unit 4.

  FIG. 2 is a functional block diagram illustrating a functional group realized by the information processing apparatus 2. In the information processing apparatus 2, a call voice data storage unit 8a, a call summary storage unit 8b, and an important word storage unit 8c are realized. These are realized by the hard disk 8.

  In the information processing apparatus 2, a keyword storage unit 6a, a word spotting result storage unit 6b, and a pair storage unit 6c are further realized. These are realized by the main memory 6.

  In the information processing apparatus 2, the keyword extraction unit 4a, the speech recognition unit 4b, the word spotting unit 4c, the association execution unit 4d, the important utterance speech specification unit 4e, the first target text specification unit 4f, and the second target text specification are further provided. A unit 4g, an important word presence / absence determination unit 4h, and a call summary display unit 4f are realized. These are realized by the control unit 4 executing information processing according to the program when a call summary display operation is performed by the administrator.

[Call voice data storage]
The call voice data storage unit 8a stores call voice data 14 which is voice data indicating a series of uttered voices made by an operator working in a manufacturer's call center in a call with a customer. The call voice data 14 is voice recording data input to the telephone receiver by the operator during a call.

  FIG. 3 is a diagram illustrating an example of the call voice data 14. Arrows indicate the passage of time since the start of the call. The call voice data 14 includes a voice portion related to a series of uttered voices uttered by an operator. The part where the operator is not speaking is hatched. The call voice data 14 records the start timing and end timing of each utterance voice. In the present embodiment, a serial number (hereinafter referred to as an utterance number) is assigned to each utterance voice from the beginning, and the utterance number of each utterance voice is recorded in the call voice data 14. FIG. 4 shows an example of a part of speech voice. In FIG. 4, the utterance number is written to the left of each utterance voice.

[Call summary storage]
The call summary storage unit 8b stores call summary data. The call summary data is document data indicating a text document created by the operator while recalling the contents of the call after the call with the customer is finished. The contents of the call are recorded in the text document, and the text of each of a plurality of sentences (hereinafter referred to as text sentences) is included. That is, the text document includes a plurality of text sentences (corresponding to a plurality of unit texts) relating to the utterance part of the operator and a plurality of text sentences relating to the utterance part of the customer. In the case of the present embodiment, the document data includes information indicating which text sentence is the text sentence related to the utterance portion of the operator or the customer. FIG. 5 shows an example of the contents of a text document. In the case of this embodiment, the character string described in one line becomes one text sentence. In FIG. 5, for convenience, a serial number is written on the left of each text sentence. Further, in FIG. 5, for convenience, a text sentence related to the utterance portion of the operator is shown in bold.

  Hereinafter, a text sentence related to the utterance part of the operator is referred to as a target text.

[Key word storage]
The important word storage unit 8c stores an important word list. The important word list is data indicating a plurality of important words registered in advance by the administrator. The important word list also indicates how to read each important word. FIG. 6 is a diagram illustrating an example of the important word list. As shown in the figure, the important word list stores, for each important word, a character string of the important word and a phoneme indicating how to read the important word in association with each other.

  Note that the character strings and phonemes of each important word represented by the important word list are registered in advance in a word dictionary for speech recognition.

  The call voice data, call summary data, and important word list are read out by the control unit 4 (voice acquisition means) when the above-described call summary display operation is performed.

  Next, the keyword extraction unit 4a, the speech recognition unit 4b, the word spotting unit 4c, the association execution unit 4d, and the important utterance speech specifying unit 4e will be described.

[Keyword extractor]
The keyword extraction unit 4a performs morphological analysis processing on each target text, and extracts independent words included in each target text as keywords.

  In the present embodiment, the keyword extraction unit 4a first assigns a serial number (hereinafter referred to as a sentence number) from the first text sentence to each text sentence in the text document. Thereafter, the keyword extraction unit 4a performs morphological analysis processing for each target text in the text document to extract independent words included in the target text as keywords, and associates the extracted keyword with the sentence number of the target text. And stored in the keyword storage unit 6a. Note that MeCab, teacup, and JUMAN are used as the morphological analyzer for the morphological analysis processing. FIG. 7 shows the stored contents of the keyword storage unit 6a. The numbers in parentheses indicate sentence numbers. Moreover, the keyword extracted from the object text of the sentence number is shown on the right side of the sentence number.

[Voice recognition part]
The voice recognition unit 4b performs voice recognition processing on the call voice data. In the case of the present embodiment, the voice recognition unit 4b performs voice recognition processing on the call voice data in accordance with a voice recognition algorithm used by the open source voice recognition engine “Julius”. Thereby, the speech recognition unit 4b obtains a so-called word lattice that is a result of the speech recognition process for each uttered speech. The voice recognition unit 4 b stores the word lattice data obtained from the uttered voice in the main memory 6 in association with the utterance number of the uttered voice.

  FIG. 8 is a diagram illustrating an example of a word lattice. As shown in the figure, the word lattice represents a graph including each word and a link connecting words that can be connected. Each word corresponds to a node. A rectangle represents a node (word), and a line connecting the rectangles represents a link.

[Word spotting part]
For each utterance voice, the word spotting unit 4c specifies an important word and a keyword uttered by the utterance voice based on word lattice data associated with the utterance number of the utterance voice.

  In the present embodiment, the word spotting unit 4c first stores data indicating a set of keywords (hereinafter referred to as a keyword set) extracted from each target text by the keyword extraction unit 4a based on the storage contents of the keyword storage unit 6a. Generate. Then, for each utterance voice, the word spotting unit 4c selects an important word and keyword (that is, an important word and keyword uttered by the utterance voice) included in the word lattice associated with the utterance number of the utterance voice as an important word list. The keyword is identified based on the data indicating the keyword set, and the identified important word and keyword are stored in the word spotting result storage unit 6b in association with the utterance number of the uttered voice.

  FIG. 9 shows the stored contents of the word spotting result storage unit 6b. The numbers in parentheses indicate utterance numbers. Further, on the right side of the utterance number, important words and keywords uttered by the utterance voice of the utterance number are shown. Important words and keywords are stored separately. In FIG. 9, important words are double underlined and keywords are single underlined.

[Associate execution part]
The association execution unit 4d associates each target text with one of the uttered voices.

  FIG. 10 is a flowchart showing processing executed by the association execution unit 4d. First, the association executing unit 4d sorts a plurality of target texts included in the text document in ascending order of sentence numbers (S101). Hereinafter, the “i” th target text is referred to as target text [i].

  Further, the association execution unit 4d sets a speech voice set (S102). That is, in S102, the association execution unit 4d sets all of the plurality of utterance voices represented by the call voice data as the source of the utterance voice set.

  Then, the steps after S103 are executed in order from the first target text [1].

  That is, the association execution unit 4d sorts the speech sounds included in the speech sound set in ascending order of the speech numbers (S103). Hereinafter, the “j” -th utterance voice is denoted as utterance voice [j]. Then, the association execution unit 4d executes steps S104 and S105 in order from the first utterance voice [1].

  That is, the association execution unit 4d counts the number of keywords (hereinafter referred to as the number of keywords) shared between the target text [i] and the uttered voice [j] (S104). More specifically, in S104, the association execution unit 4d associates the keyword spotting result with the set of keywords stored in the keyword storage unit 6a in association with the sentence number of the target text [i] and the speech number of the utterance voice [j]. The number of keywords included in both the keyword set stored in the storage unit 6b is counted as the number of keywords.

  The association executing unit 4d stores the number of keywords counted in S104 in the main memory 6 in association with the utterance number of the uttered voice [j]. Thus, the number of keywords is counted for all uttered voices in the uttered voice set. Thereafter, the association execution unit 4d identifies the utterance number associated with the maximum number of keywords stored in the main memory 6 (S106), and the identified utterance number and the sentence number of the target text [i]. Is stored in the pair storage unit 6c (S107). If all the keywords stored in the main memory 6 are “0”, S106 and S107 are skipped.

  Further, the association executing unit 4d deletes the utterance voice to which the utterance number before the utterance number specified in step S106 is assigned from the utterance voice set (S108).

  FIG. 11 is a conceptual diagram showing a state in which the target text is associated with the uttered voice by the association execution unit 4d. Each arrow indicates the speech to be linked. According to the figure, the target text with the sentence number “24” is associated with the utterance voice with the utterance number “34”. Further, the target text with the sentence number “25” is associated with the utterance voice with the utterance number “38”. The target text with the sentence number “26” is associated with the utterance voice with the utterance number “44”.

[Important speech identification unit]
The important utterance voice specifying unit 4e specifies an important utterance voice that is a utterance voice including the voice of any one of the important words among the plurality of utterance voices represented by the call voice data. In the case of the present embodiment, the important utterance voice specifying unit 4e specifies the important utterance voice based on the stored contents of the word spotting result storage unit 6b. More specifically, the important utterance voice specifying unit 4e specifies one or a plurality of utterance numbers associated with any one of the utterance numbers stored in the word spotting result storage unit 6b.

  Next, the first target text specifying unit 4f, the second target text specifying unit 4g, the important word presence / absence determining unit 4h, and the call summary display unit 4f will be described. Hereinafter, an utterance voice of an arbitrary utterance number among one or a plurality of utterance numbers specified by the important utterance voice specifying unit 4e is referred to as an important utterance voice X.

[First target text identification part]
The first target text specifying unit 4f specifies the target text (hereinafter referred to as the first target text) associated with the utterance voice whose utterance number is before the important utterance voice X. Specifically, the first target text specifying unit 4f specifies a pair including the largest utterance number among pairs including an utterance number smaller than the utterance number of the important utterance voice X, and a sentence included in the specified pair. The number is specified as the sentence number of the first target text. For example, when the utterance voice of the utterance number “41” shown in FIG. 11 is the important utterance voice X, the target text of the sentence number “25” associated with the utterance voice of the utterance number “38” is the first target text. Identified.

[Second target text identification part]
The second target text specifying unit 4g specifies a target text (hereinafter referred to as a second target text) associated with an utterance voice whose utterance number is later than the important utterance voice X. Specifically, the second target text specifying unit 4g specifies a pair including the smallest utterance number among pairs including an utterance number larger than the utterance number of the important utterance voice X, and a sentence included in the specified pair. The number is specified as the sentence number of the second target text. For example, when the utterance voice of the utterance number “41” shown in FIG. 11 is the important utterance voice X, the target text of the sentence number “26” associated with the utterance voice of the utterance number “44” is the second target text. Identified.

[Important word presence determination unit]
The important word presence / absence determination unit 4h determines whether or not there is a target text including a character string of an important word uttered by the important utterance speech X between the first target text and the second target text.

  FIG. 12 is a flowchart showing processing executed by the important word presence / absence determination unit 4h. First, the important word presence / absence determination unit 4h sets the value of the flag stored in the main memory 6 to “0” (S201). Further, the important word presence / absence determination unit 4h determines whether or not the sentence number of the first target text and the sentence number of the second target text are serial numbers (S202). When the sentence number of the first target text and the sentence number of the second target text are serial numbers (YES in S202), the important word presence / absence determination unit 4h ends the process.

  On the other hand, when the sentence number of the first target text and the sentence number of the second target text are not consecutive numbers (NO in S202), the important word presence / absence determining unit 4h determines the sentence number of the first target text and the second number. The target texts with the sentence numbers between the sentence numbers of the target texts are sorted in ascending order of the sentence numbers (S203). Hereinafter, the “i” th target text is referred to as target text [i].

  Then, the important word presence / absence determination unit 4h stores the target text [i] in the word spotting result storage unit 6b in association with the utterance number of the important utterance speech X in order from the “1” -th target text [1]. It is determined whether or not an important word character string is included based on the call summary data and the important word list (S204). When the target text [i] includes a character string of an important word stored in association with the utterance number of the important utterance voice X (YES in S204), the important word presence / absence determination unit 4h sets the flag value to “1”. Update (S205) and end the process.

  The flag value “0” indicates that “the target text including the character string of the important word uttered by the important utterance speech X does not exist between the first target text and the second target text”. “1” indicates that “the target text including the character string of the important word uttered by the important speech X exists between the first target text and the second target text”.

[Call summary display]
The call summary display unit 4 i displays an image 16 of the text document represented by the call summary data on the display unit 10. However, when the value of the flag is “0”, that is, when the target text including the character string of the important word uttered by the important utterance speech X does not exist between the first target text and the second target text. The call summary display unit 4i outputs information indicating that. That is, when the value of the flag is “0”, the call summary display unit 4i, as shown in FIG. 13, shows information (in this case, a character string) indicating an important word associated with the utterance number of the important utterance voice X. ) Including images, icons, windows, and the like are displayed on the display unit 10 together with the image 16. The character string “Fujisan x430” indicates a character string of an important word.

  As described above, in the information processing apparatus 2, it is determined whether or not the important word issued by the operator in the call is described in an appropriate place in the text document. In addition, the manager is notified that the important word issued by the operator in the call is not recorded in an appropriate place in the text document.

  In addition, embodiment of this invention is not restricted only to the said embodiment.

[Modification]
For example, the important word presence / absence determining unit 4h also determines whether or not there is a target text including an alternative character string having a predetermined relationship with the character string of the important word between the first target text and the second target text. Good. Here, the substitute character string is, for example, a character string of a broader concept word or a lower concept word of a key word, a character string of a synonym of a key word, and a case of a part of a character string of a key word (uppercase or lowercase letters). A character string with a changed distinction. Here, a case where the alternative character string is “a character string obtained by changing the case of a part of a character string of an important word” will be described as an example, and a modified example will be described.

  FIG. 14 is a flowchart showing processing executed by the important word presence / absence determination unit 4h in the modification. As shown in the figure, in the modification, steps S206 and S207 are added to the process shown in FIG. That is, the important word presence / absence determination unit 4h further determines that the target text [i] does not include the important word character string stored in association with the utterance number of the important utterance voice X (NO in S205). It is determined whether i] includes an alternative character string (S206). When the target text [i] includes an alternative character string (YES in S206), the important word presence / absence determining unit 4h stores the sentence number of the target text [i] in the main memory 6 as the sentence number X, and The value of the flag is updated to “2” (S207), and the process ends.

  The flag value “2” indicates that “the target text including the substitute character string exists between the first target text and the second target text”.

  Further, in the modification, when the call summary display unit 4i has the flag value “2”, as shown in FIG. 15, the important word associated with the utterance number of the important utterance voice X and its substitute character string A correction guide image 20 such as a figure, an icon, and a window including information (here, a character string) indicating “” is displayed on the display unit 10 together with the image 16. The character string “Fujisan x430” is a character string of an important word, and the character string “FUJISAN X430” indicates an alternative character string. As shown in FIG. 15, the correction guide image 20 includes a correction instruction image 22 and a cancel instruction image 24.

  In the modified example, when an operation for selecting the correction instruction button image 22 is performed, the call summary display unit 4i updates the call summary data as described below. That is, the call summary display unit 4i corrects the substitute character string included in the target text of the sentence number X to the important word character string associated with the utterance number of the important utterance voice X.

  2 Information processing device, 4 control unit, 4a keyword extraction unit, 4b speech recognition unit, 4c word spotting unit, 4d association execution unit, 4e important utterance speech specification unit, 4f first target text specification unit, 4g second target text specification 4h important word presence / absence determination unit, 4i call summary display unit, 6 main memory, 6a keyword storage unit, 6b word spotting result storage unit, 6c pair storage unit, 8 hard disk, 8a call voice data storage unit, 8b call summary storage Part, 8c important word storage part, 10 display part, 12 operation input part, 14 call voice data, 16, 18 images, 20 correction guidance images, 22 correction instruction images, 24 cancel instruction images.

Claims (8)

Based on the result of the morphological analysis process for each of the plurality of unit texts included in the text document in which the content of the call is recorded and the result of the speech recognition process for each of the plurality of uttered voices uttered in the call indicated by the voice data , Means for associating each unit text with one of the utterances,
A first text that is a unit text associated with an utterance voice prior to an important utterance voice that is an utterance voice including a voice of a predetermined important word among the plurality of utterance voices; A determination means for determining whether or not there is a unit text including the character string of the important word between the second text that is a unit text associated with the uttered voice;
An information processing apparatus including:   2. When it is determined that there is no unit text including the key word character string between the first text and the second text, a message to that effect is output. Information processing device Further comprising display control means for displaying the text document on display means,
The display control means includes
When there is no unit text including the important word character string between the first text and the second text, the information indicating the important word is displayed on the display unit together with the text document.
The information processing apparatus according to claim 2. Means for determining whether or not there is a unit text including an alternative character string that is a character string having a predetermined relationship with a character string of the important word between the first text and the second text; Including
The display means includes
When there is a unit text including the substitute character string between the first text and the second text, information indicating the important word and the substitute character string is displayed on the display unit together with the text document. Letting
The information processing apparatus according to claim 3. When there is a unit text including the substitution character string between the first text and the second text, the text document is updated, and the unit text existing between the first text and the second text Means for correcting the substitute character string included in the key word character string;
The information processing apparatus according to claim 4. An independent word extraction means for extracting independent words included in each unit text by morphological analysis processing;
A self-supporting word specifying means for specifying a self-supporting word extracted by the self-supporting word extraction means, which is uttered by each utterance voice by a speech recognition process;
Further including
The association means includes
Associating each unit text with an utterance that produces the same independent word as that contained in the unit text;
The information processing apparatus according to any one of claims 1 to 5.   The information processing apparatus according to claim 1, further comprising a voice acquisition unit that acquires the voice data indicating the plurality of uttered voices. A result of a morphological analysis process for each of a plurality of unit texts included in a text document in which the contents of the call are recorded, and a result of a speech recognition process for each of a plurality of speech sounds uttered in the call indicated by voice data. An association means for associating each unit text with one of the utterances,
A first text that is a unit text associated with an utterance voice prior to an important utterance voice that is an utterance voice including a voice of a predetermined important word among the plurality of utterance voices; Determination means for determining whether or not there is a unit text including the character string of the important word between a second text which is a unit text associated with the uttered voice;
As a program to make the computer function as.

Images