JP6281856B2 - Local language resource reinforcement device and service providing equipment device - Google Patents

Description

  The present invention relates to a service using a word dictionary, such as speech recognition and automatic translation, and more particularly to a technique for reinforcing a user dictionary in order to improve the accuracy of speech recognition or automatic translation.

  With the spread of portable telephones, particularly so-called smartphones, various applications have appeared. In particular, applications that use speech recognition for input are expected to become more popular. This is because it is difficult to input text with a small device such as a smartphone.

  However, in order to further spread voice recognition, it is necessary to further improve the accuracy of voice recognition. One way to increase accuracy is to enrich the dictionary used for speech recognition. This is because in speech recognition, it is difficult in principle to recognize words that are not in the dictionary. Even now, not only speech recognition but also a system for processing data related to speech generally includes a dictionary having tens of thousands to hundreds of thousands of vocabularies.

  On the other hand, infrequently used words such as technical terms, new words, and buzzwords are often not registered in these dictionaries. If speech containing such vocabulary is input to the system, results of appropriate speech processing cannot be obtained.

  In order to deal with such a problem, such a system is generally provided with a user dictionary in which a user can register a vocabulary. By registering vocabulary frequently used by the user in the user dictionary, it is possible to improve the accuracy of voice processing while suppressing an increase in the processing amount.

  A similar problem occurs with automatic translation. Even in automatic translation, by providing a user dictionary for a translation dictionary consisting of a pair of vocabularies in two languages and a dictionary of vocabularies in the target language, the translation accuracy of a sentence related to a user's specialized field can be improved. Can be increased.

  However, there is a problem that the user dictionary cannot be effectively used at present. One of the reasons is that the procedure for registering the vocabulary in the user dictionary is complicated. Some users make effective use of user dictionaries. However, in order for general users to use user dictionaries, it is possible to reinforce user dictionaries more easily, or user dictionaries can be registered by users. It is necessary to be able to reinforce the user dictionary without the work of.

  An invention for solving such a problem is proposed in Patent Document 1 described later. The speech recognition terminal of the speech recognition system disclosed in Patent Document 1 basically performs speech recognition using a speech recognition dictionary provided in the speech recognition terminal. If this voice recognition fails, the voice recognition terminal transmits the voice data to the voice recognition server. The voice recognition server performs voice recognition using a voice recognition dictionary having a vocabulary much larger than the dictionary of the voice recognition terminal, and transmits the result to the voice recognition terminal. The speech recognition result word is registered in the speech recognition dictionary together with the original speech data. Accordingly, words that have failed to be recognized by the speech recognition terminal (normally words that do not exist in the dictionary of the speech recognition terminal) are additionally registered in the dictionary of the speech recognition terminal. According to the disclosure of Patent Document 1, a user does not intervene in the processing during this time, and a new word is easily registered in the dictionary of the speech recognition terminal.

JP 2012-88370 A

  However, the system disclosed in Patent Document 1 still has problems to be solved as follows.

  First, a vocabulary that does not appear in the user's utterance has a problem that a word cannot be registered in the dictionary of the speech recognition terminal. Patent Document 1 does not mention at all about registering a vocabulary that does not appear in the utterance by the user in the dictionary of the terminal.

  Second, the invention described in Patent Document 1 does not matter at all about the user's utterance content. If the user's utterances tend to concentrate on a specific field, vocabulary other than that specific field is not very effective even if it is registered in the user dictionary. This is the same even when the utterance is processed by the server. Patent Document 1 shows no interest in reinforcing the user dictionary in such a way as to suit a specific field.

  Therefore, an object of the present invention is to provide a local operation provided for each terminal in a manner suitable for the user's language processing requirements in a manner that does not reduce the accuracy of the language processing for each terminal. A language resource reinforcement device that can reinforce a local language resource used for language processing, and a service providing facility device that can realize reinforcement of a local language resource by providing a language service by communicating with such a language resource reinforcement device Is to provide.

  A local language resource reinforcement device according to a first aspect of the present invention is a service that uses a language service by communicating online with a service providing equipment device that provides a language service using a language resource prepared in advance. Providing services when it is possible to communicate with the use means, local service execution means for locally executing a service of the same type as the language service using a local language resource having a smaller scale than the language resource, and the service providing equipment. Language resource reinforcing means for reinforcing the local language resource using the language resource for reinforcement transmitted from the facility device.

  Preferably, the local language resource reinforcement device further includes a reinforcement request means for transmitting a reinforcement request for requesting reinforcement of the local language resource to the service providing equipment device when communication with the service providing equipment device is possible. Including. The language resource reinforcing means includes means for reinforcing the local language resource using the reinforcing language resource transmitted from the service providing facility device in response to the reinforcement request.

  More preferably, the reinforcement requesting means is a processing requesting means for requesting the service providing facility apparatus to perform processing by the language service for the language data together with the language data to be processed when the communication is possible with the service providing facility apparatus. including. In response to the processing request from the processing requesting means, the service providing equipment device has processed the language data to be processed by the language service, and the language resource used in the process of obtaining the processing result of the language resource. The language resource for reinforcement consisting of is transmitted to the local language resource reinforcement device together or separately. The means for reinforcing is for receiving the language resource for reinforcement transmitted from the service providing equipment when the communication is possible with the service providing equipment and adding the language resource for reinforcement to the local language resource. Language resource addition means.

  More preferably, the reinforcement requesting means is a local language resource transmitting means for transmitting a reinforcement request for requesting reinforcement of the local language resource together with the local language resource to the service providing equipment when communication with the service providing equipment is possible. including. The means for reinforcing is for replacing the local language resource before reinforcement with the local language resource after reinforcement transmitted from the service providing facility device in response to the reinforcement request transmitted by the local language resource transmission means. Local language resource replacement means may be included.

  A service providing facility apparatus according to a second aspect of the present invention provides a service providing means for providing a language service using a language resource prepared in advance to another client device via online communication, and a service providing Communicating with a reinforcement language resource creating means for creating a reinforcement language resource to reinforce a local language resource used locally in another terminal apparatus by using a service execution result by the means, and another terminal apparatus When possible, it includes reinforcement language resource transmission means for transmitting the reinforcement language resource created by the reinforcement language resource creation means to another terminal device.

  Preferably, the reinforcing language resource creating means receives the reinforcement request for requesting reinforcement of the local language resource used locally in the other terminal device from the other terminal device, and the service providing means provides the service. Means for creating a supplemental language resource using the partial language resource used in the provision;

  More preferably, the reinforcing language resource creating means is a partial language resource used when the service providing means provides a service in response to a request from another terminal device among the prepared language resources. Temporary storage means for temporarily storing. The reinforcement language resource transmitting means transmits the partial language resources stored in the temporary storage means to another terminal device together with or separately from the processing result of the service provided by the service providing means. Transmitting means.

  Preferably, the local language resource includes one or more data items that are distinguished from each other. The reinforcing language resource creating means includes a local language resource storage means for collecting and storing local language resources used in each terminal device from a plurality of terminal devices, and a local language resource stored in the local language resource storage means. And a frequent data item group specifying means for specifying one or a plurality of frequent data item groups that are determined to have a higher possibility of co-occurrence than a threshold, and for each of the plurality of terminal devices, 1 or A frequent data item group in which only some of the data items are included in the local language resource transmitted from the terminal device is identified from among the multiple frequent data item groups, and based on the identified frequent data item group Means for creating a reinforcing language resource for the terminal device.

  As described above, according to the present invention, the terminal device executes locally the same kind of application as the language service provided online by the service providing facility device, using the local language resource having a smaller scale than the language resource. The reinforcement request means transmits a reinforcement request for requesting reinforcement of the local language resource to the service providing equipment device when the terminal device can communicate with the service providing equipment. Upon receiving this request for reinforcement, the service providing facility device creates a language resource for reinforcement that is likely to be used in the terminal device, and transmits it to the terminal device. The language resource reinforcement means of the terminal device reinforces the local language resource using the language resource for reinforcement.

  In the terminal device, in order to input and output the language resource for reinforcement, it is only necessary to communicate with the service providing facility device and use the service. Therefore, a particularly complicated operation is not required. From the service providing equipment, for example, the data obtained during the provision of the language service, or data mining for the language resources of other users and the user's language resources of the terminal device, may be used in the terminal device A data item of a high language resource is created as a reinforcing language resource and transmitted to the terminal device.

  As a result, for each terminal, a local language resource reinforcement device that can reinforce local language resources in a form that conforms to the requirements of language processing by the user in a manner that does not reduce the accuracy of speech recognition with a simple operation, In addition, by providing a language service by communicating with such a local language resource reinforcement apparatus, it is possible to provide a service providing facility apparatus that can realize the reinforcement of the local language resource.

It is a figure which shows typically the whole structure of the speech translation system which concerns on the 1st Embodiment of this invention. It is a sequence diagram which shows typically the communication sequence between the portable terminal and the speech translation server in the system shown in FIG. It is a block diagram which shows the hardware constitutions of the portable terminal used with the system of 1st Embodiment. It is a block diagram which shows the hardware constitutions of the computer system which implement | achieves the speech translation server used with the system of 1st Embodiment. 5 is a flowchart of a program for providing a speech translation service in response to a speech translation request from a portable terminal in the speech translation server of the system according to the first embodiment. 4 is a flowchart showing a control structure of a program for registering a vocabulary for a personal dictionary in a personal dictionary transmitted from a speech translation server in a portable terminal of the system according to the first embodiment. The sequence which shows typically the communication sequence performed in order to register or update a user supplement dictionary performed between a portable terminal and a speech translation server in the speech translation server concerning a 2nd embodiment of the present invention. FIG. It is a flowchart which shows the control structure of the program which produces the personal dictionary vocabulary for each user in the speech translation server of the system which concerns on 2nd Embodiment.

  In the following description and drawings, the same parts are denoted by the same reference numerals. Therefore, detailed description thereof will not be repeated.

[First Embodiment]
<Outline>
<< Overall configuration (Fig. 1) >>
Referring to FIG. 1, a speech translation system 100 according to the present invention provides a speech translation service that is a kind of language service in response to a speech translation request from a speech translation client device connected to the Internet 102. If it is possible to connect to the server 106, which is a facility device, and the Internet 102 and communicate with the server 106 via the Internet 102, the speech translation service provided by the server 106 is used (FIG. 1 (A)). When the mobile terminal 104 cannot communicate with the server 106 (FIG. 1B), the portable terminal 104 having a function of executing a speech translation service (hereinafter referred to as “local speech translation service”) 112 in a local environment. Including. The local speech translation service 112 can be considered as a function that provides a speech translation service locally when communication with the server 106 is not possible online.

  The server 106 performs a speech translation service using a large-capacity speech processing dictionary (large vocabulary dictionary 108). The large vocabulary dictionary 108 stores a large number of data items (dictionary entries) representing individual vocabularies in a form that can be distinguished from each other. In the speech translation service, in addition to the acoustic model that is a statistical model for speech recognition and the translation model that is a statistical model for automatic translation, in general, a word dictionary for speech recognition of the source language, A target language word dictionary at the time of translation, a translation dictionary corresponding to the combination of the source language and the target language, and a word dictionary at the time of speech synthesis of the target language are required. In the following description, a case will be described in which the server 106 performs processing for reinforcing the user dictionary for each terminal (or for each user) mainly for a word dictionary and a translation dictionary for speech recognition and speech synthesis.

  In FIG. 1A, when the portable terminal 104 uses the speech translation service of the server 106, the following procedure is performed. First, the voice uttered by the user is sent from the portable terminal 104 to the server 106 together with a translation request. The server 106 performs speech recognition on the speech using the large vocabulary dictionary 108 and outputs a source language morpheme string. Further, the server 106 receives this morpheme string as an input and performs automatic translation according to the information of the language pair attached to the translation request. At this time, the language dictionary of the target language in the large vocabulary dictionary 108 and the translation dictionary of the designated language pair are used. As a result of automatic translation, a morpheme string in the target language is obtained. Therefore, speech synthesis is performed on the morpheme string, and the synthesized speech is transmitted from the server 106 to the portable terminal 104. In this speech synthesis, the target language speech synthesis word dictionary in the large vocabulary dictionary 108 is used. The portable terminal 104 utters this synthesized voice through a speaker.

  Referring to FIG. 1B, the portable terminal 104 incorporates a local dictionary 110 for speech processing that is smaller than the large vocabulary dictionary 108 used to execute the local speech translation service 112. The local dictionary 110 stores a number of data items smaller than the number of data items in the large vocabulary dictionary 108, with vocabulary distinguished from each other as individual data items. The reason why this number is smaller than the number of data items in the large vocabulary dictionary 108 is due to restrictions on hardware and performance of the portable terminal 104.

  In the embodiment of FIG. 1B, in the above-described speech translation process, the server 106 extracts word candidates to be additionally registered in the local dictionary 110 of the portable terminal 104, temporarily stores them, and then performs speech translation. When the result is transmitted, it is sent to the portable terminal 104 together. In the portable terminal 104, only the received word candidates that do not exist in the local dictionary 110 are additionally registered in the local dictionary 110. Details of these processes will be described later.

  Each entry in the speech recognition dictionary includes a word (or morpheme) that is a headword and information necessary for functioning voice recognition for the headword.

  Each entry in the translation dictionary includes a word (or morpheme) that is a headword and information necessary for the language translation for the headword to function.

  Each entry of the dictionary for speech synthesis includes a word (or morpheme) that is a headword and information necessary for functioning speech synthesis for the headword.

─ Speech translation and dictionary registration sequence (Fig. 2) ─
With reference to FIG. 2, a typical communication sequence between the portable terminal 104 and the server 106 at the time of speech translation using the speech translation service provided by the server 106 of the speech translation system 100 will be described. Here, the local dictionary 110, which is a language resource locally held by the portable terminal 104, is a basic dictionary containing basic vocabulary used by the local speech translation service 112 operating locally on the portable terminal 104. 134 and a user dictionary 130 in which a user can register vocabulary independently. The local dictionary 110 according to the present embodiment is a word that is not included in the basic dictionary 134 and not included in the user dictionary 130 by the function of the server 106, and is used in the process of the speech translation service performed by the server 106. A personal dictionary 132 for reinforcing the user dictionary 130, which includes word dictionary entries for speech recognition, automatic translation, and speech synthesis temporarily stored in the storage device, is included.

  First, voice input 150 is performed in portable terminal 104, and a voice recognition request including the voice data and information such as a language combination of voice translation is transmitted to server 106. In this embodiment, this request is a request for speech recognition, and is also a reinforcement request made to the server 106 to perform a process for reinforcing the local user dictionary 130.

  When the server 106 receives the voice recognition request, the server 106 performs voice recognition processing 152. In the process of speech recognition, the server 106 temporarily records the dictionary entry used by the speech recognition process 152 in the large vocabulary dictionary 108 in the personal dictionary creation storage unit 380 (see FIG. 4). Subsequently, the server 106 performs automatic translation 154 to the target language on the morpheme string of the source language output by the speech recognition processing 152. Even during the automatic translation process, the server 106 temporarily records the entry used in the automatic translation 154 of the speech recognition dictionary in the large vocabulary dictionary 108 in the personal dictionary creation storage unit 380. Furthermore, the server 106 performs speech synthesis 156 using speech synthesis resources (not shown) based on the target language morpheme sequence obtained by the automatic translation 154. Even during the speech synthesis process, the server 106 temporarily stores the entry used in the speech synthesis 156 of the speech synthesis dictionary in the large vocabulary dictionary 108 in the personal dictionary creation storage unit 380. Finally, the server 106 is a language resource for reinforcing the personal dictionary 132 temporarily recorded in the personal dictionary creation storage unit 380, and the service use result including the speech recognition result, the automatic translation result, and the synthesized speech. The dictionary entry is transmitted to the portable terminal 104 (process 158). The server 106 back-translates the translation result by the automatic translation 154 into the source language, and transmits the resulting morpheme string to the portable terminal 104. By displaying the reverse translation on the portable terminal 104 in this way, the user can determine whether or not the content intended by the user has been translated into the target language. However, since reverse translation is not directly related to the present invention, reverse translation is not described in the following description.

  The portable terminal 104 that has received the speech recognition result, automatic translation result, synthesized speech, and dictionary entry from the server 106 displays the text of the result of speech recognition and automatic translation on the screen, and simultaneously uses the synthesized speech using a speaker. To speak (process 160). By making this voice heard by the other party who is interacting with the user of the portable terminal 104, the user can interact with the other party.

  Further, in the subsequent process 162, the speech recognition and speech synthesis dictionary entries of the source language and the target language received from the server 106 and the translation dictionary entries corresponding to the combination of the source language and the target language are stored in the personal dictionary 132. It is added to the dictionary for speech recognition and speech synthesis of the source language and the target language, and the translation dictionary corresponding to the combination of the source language and the target language. This process reinforces the user dictionary 130 which is a local language resource. However, words already stored in the user dictionary 130 or the basic dictionary 134 among the dictionary entries transmitted from the server 106 are not added to the personal dictionary 132.

  The creation and temporary storage of the dictionary entry sent by the server 106 and the transmission to the client device do not have to be performed at the same time as the provision of the speech translation service, and may be at any timing. For example, a speech translation history may be recorded for each client device, and dictionary entries may be created and stored based on the history. The entry need not be transmitted simultaneously with the provision of the speech translation service, and may be stored in the server 106 and transmitted to the client device at some timing. If the vocabulary of the basic dictionary 134 of the portable terminal 104 is known in the server 106, the entries installed in the basic dictionary are excluded from the dictionary entries transmitted to the portable terminal 104 in the process 158. Also good.

  FIG. 2 shows a typical processing sequence. It goes without saying that even if this sequence is different from this processing sequence, the same effect as this embodiment can be obtained as long as it is a processing sequence that includes essentially the same communication as that shown in FIG. 2 in the same order. Absent. When the user further uses the speech translation service, the sequence of FIG. 2 is repeatedly executed.

  Note that the client device may have a function of automatically maintaining entries registered in the personal dictionary according to the frequency with which each entry is used in the client device. For example, the number of times used in the speech translation service may be stored for each entry in the personal dictionary, and when the personal dictionary is enlarged and exceeds a certain capacity, an entry with low usage frequency may be deleted. In this way, the capacity of the personal dictionary can be maintained below a certain level.

<Hardware configuration>
<Portable terminal 104 (FIG. 3)>
With reference to FIG. 3, the portable terminal 104 executes a predetermined program and controls each unit of the portable terminal 104, thereby implementing a processor 250 for realizing various functions, and various types executed by the processor 250. A memory 252 for storing programs and data, a RAM 262 for providing a working storage area for the processor 250, and an interface 254 between the processor 250 and various sensors described later are included. Any of the components described below can communicate with the processor 250 via the interface 254.

  The portable terminal 104 further includes a microphone 256, a GPS receiver 258 for acquiring longitude and latitude information of the position of the portable terminal 104 by the GPS function, various sensor groups 260, and a base station (not shown) by wireless communication. In addition to the communication device 272 that can be connected to the Internet 102, the touch panel 274, and the touch panel 274, an operation button 276 and a speaker 280 provided on the housing of the portable terminal 104 are included.

  The memory 252 includes a speech translation client application program (hereinafter referred to as “speech translation client”) 300 for using a speech translation service provided by the server 106 when it can communicate with the server 106, and the local speech translation service 112. (See FIG. 1) Application group, acoustic model and translation model, local program group 302 such as acoustic resources for speech synthesis, dictionary group 304 corresponding to the above-mentioned local dictionary 110, and server A personal dictionary automatic registration program 306 that executes processing for adding an entry to the personal dictionary 132 through communication with the user dictionary 106 and a dictionary maintenance program 308 that is used by the user to maintain the user dictionary 130 are stored.

<< Server 106 (FIG. 4) >>
The server 106 according to the above embodiment can be realized by computer hardware and a computer program executed on the computer hardware. FIG. 4 shows the internal configuration of the computer constituting the server 106.

  Referring to FIG. 4, the server 106 includes a LAN 378 connected to the Internet 102 via a router 376 and a computer 340 connected to the LAN 378 and having a memory port 352 and a DVD (Digital Versatile Disc) drive 350. A keyboard 346, a mouse 348, and a monitor 342 connected to the computer 340 are included. Server 106 may further include a printer connected to computer 340.

  The computer 340 includes a CPU (Central Processing Unit) 356 and a bus 366 connected to the CPU 356 in addition to the memory port 352 and the DVD drive 350. A memory port 352 and a DVD drive 350 are also connected to the bus 366. The computer 340 is further connected to a bus 366, and a read only memory (ROM) 358 that stores a boot-up program and the like, and a RAM 360 that is connected to the bus 366 and temporarily stores a program command, a system program, work data, and the like. Including. The server 106 further uses a program for speech recognition, automatic translation, speech synthesis, etc. used by the CPU 356, and various data such as an acoustic model, translation model, word dictionary, speech unit for speech synthesis, and the like necessary for these programs. A hard disk 354 that stores an operating system program of the computer 340, and a network interface card that provides the computer 340 with connection to another terminal (eg, the portable terminal 104) on the Internet 102 via the LAN 378 or the router 376 NIC) 368 and a dictionary entry that is referred to and used for processing of a speech translation request from a user among various dictionaries stored in HDD 354 and temporarily associated with the terminal ID of portable terminal 104. Personal computer consisting of hard disk And a book created for the storage unit 380.

  The hard disk 354 is a non-volatile auxiliary storage device that stores data such as a computer program, an operating system (OS), and various models for realizing each functional unit of the speech translation server by the computer hardware of the server 106. This computer program is stored in the DVD 362 or the removable memory 364 mounted in the DVD drive 350 or the memory port 352, and further transferred to and stored in the hard disk 354. Alternatively, the program may be transmitted to the computer 340 through the Internet 102, the router 376 and the NIC 368 and stored in the hard disk 354. Programs and various data for realizing the devices and methods of the above-described embodiments are appropriately loaded into the RAM 360 at the time of execution. Various programs and data may be loaded into the RAM 360 directly from the DVD 362, the removable memory 364, or via a network.

<Software configuration>
<< Server 106 (FIG. 5) >>
FIG. 5 shows a control structure of a speech translation service program executed by the server 106. The program for realizing each function of the speech translation service has the following control structure. This program includes a plurality of instructions for causing the computer 340 to function as each functional unit of the speech translation server according to the above embodiment. Some of the basic functions necessary to perform this operation are provided by an OS or a third-party program running on the computer 340, or a module or framework of various programming toolkits installed on the computer 340. . Therefore, this program does not necessarily include all the instructions necessary for realizing the system and method of this embodiment. This program realizes the above-described system function by calling an appropriate function or an appropriate program tool in a programming tool kit in a controlled manner so as to obtain a desired result according to the contents of the instruction. It only needs to contain instructions. As described above, the operation of the server 106 when a necessary instruction or a set of a series of instructions is read from the storage device and executed as necessary is well known. Therefore, detailed description thereof will not be repeated here.

  Referring to FIG. 5, after starting up this program, step 400 for initializing storage areas, initializing values of predetermined variables, confirmation of communication status, etc., and initialization of step 400 are completed. Thereafter, the process includes a step 402 of waiting for an event such as reception of a speech translation request from the terminal and branching the flow of control according to the type of the event when the event occurs. Events that may be received in step 402 may include various events such as a speech translation request from the portable terminal 104, a speech recognition result correction request, a speech translation condition setting request, and the like. Here, the processing part executed by the server 106 in response to a speech translation request that is particularly relevant to the present invention will be described. The processing when there is another request is collectively shown as step 430 in FIG. Details are not described.

  When a speech translation request is received from a client device such as the portable terminal 104 in step 402, control proceeds to step 404. The speech translation request includes speech data to be translated, information indicating a combination of translation languages, a terminal ID, its address, and the like. In step 404, it is determined whether this request is to open a new session with the client device. In the case of a new session, the server 106 generates a unique session ID, and stores the session ID and the terminal ID of the client device in the RAM 360 (FIG. 4) (step 406). Thereafter, this session ID is always used for communication with the client device to distinguish the client device. By associating the session ID with the terminal ID, the server 106 can manage processing corresponding to each terminal. Session management itself is a well-known technique, and details of session management will not be given in the following description in order to make the description and drawings easy to understand.

  Note that “new” here means that there has been no speech translation request or the like from this terminal ID within a certain past period. Since the session ID and the terminal ID are automatically discarded when there is no communication from the terminal for a predetermined period, when a speech translation request is subsequently received from the same terminal, it is treated as a new request.

  Thereafter, in step 408, voice recognition is performed on the voice data received together with the request by executing a voice recognition program. In the speech recognition process, a source language word dictionary is used to obtain a vocabulary of speech recognition results. When the speech recognition is completed, a text composed of a recognition result morpheme sequence is obtained.

  In step 410 following step 408, the voice data, the recognition result, and the dictionary entry used in the voice recognition are temporarily stored in the personal dictionary creation storage unit 380.

  In the following step 412, automatic translation into the language specified by the setting data in the speech translation request is executed on the morpheme string that is the result of speech recognition obtained in step 408. In this automatic translation process, a translation dictionary corresponding to a combination of languages and a word dictionary of the target language are used.

  In step 414 following step 412, the dictionary entry used in the automatic translation in step 412 is temporarily stored in the personal dictionary creation storage unit 380.

  In step 416 following step 414, the translation result of step 412 is back-translated into the source language, and the speech of the translation result is synthesized. Finally, a morpheme string and its associated information, a translation result, a reverse translation result, and a synthesized speech are obtained as a result of speech recognition. In step 418 following step 416, the server 106 stores these information, dictionary entries of various dictionaries used for speech recognition and automatic translation temporarily recorded in the personal dictionary creation storage unit 380, and The information indicating the type and language of the dictionary is transmitted to the terminal device that has transmitted the speech translation request, and the control returns to step 402. In most cases, a plurality of dictionary entries are transmitted for each dictionary, but all of them are transmitted to the portable terminal 104.

  If another event occurs in step 402, control branches to step 430, the process corresponding to the event is executed, and control is returned to step 402.

<Portable terminal 104 (FIG. 6)>
FIG. 6 shows a control structure of a routine portion executed when a speech translation result is received from the server 106 in the speech recognition client application executed on the portable terminal 104. Other than this part of the voice recognition client application, there is almost no relation to the present invention, so the details will not be described here.

  For example, in the present embodiment, an icon for starting this program is displayed on the touch panel screen of the portable terminal 104, and is started when the user touches the icon. On the screen, buttons for starting or ending speech input, buttons for setting a combination of speech translation, and the like are displayed. By pressing these buttons, a request corresponding to a process desired by the user is processed. Is sent to the server 106 along with the data needed to do so. FIG. 6 shows a routine executed when a speech translation result is received from the server 106 after a speech translation request is transmitted to the server 106.

  Referring to FIG. 6, this routine stores the speech recognition result text, automatic translation result text, reverse translation result text, synthesized speech data, speech recognition dictionary entry, and translation dictionary entry received from server 106. Step 450 for separating from each other, Step 452 for displaying the speech recognition result, the translation result, and the reverse translation result on the screen of the portable terminal 104, and driving the speaker 280 (FIG. 3) for the synthesized speech of the automatic translation result. And uttering step 454. By driving the speaker 280 according to the synthesized voice data, the translation result of the requested utterance is presented in the form of voice.

  Subsequently, step 456 of executing the processing described below on the plurality of dictionary entries for speech recognition received from the server 106 is executed.

  The processing executed for each of the speech recognition dictionary entries in step 456 determines whether or not there is a dictionary entry to be processed in any of the speech recognition word dictionaries in the local dictionary 110. Step 458 and step 460 of adding the dictionary entry to be processed to the voice recognition personal dictionary (part of the personal dictionary 132) when the determination in step 458 is negative. If the determination at step 458 is affirmative, nothing is done for the dictionary entry.

  The program further includes a step 462 for executing the processing described below on the plurality of dictionary entries for automatic translation received from the server 106.

  The processing executed for each entry in step 462 includes step 464 for determining whether the automatic translation dictionary entry to be processed exists in any of the local translation dictionaries in the local dictionary 110; And a step 466 of adding a dictionary entry to be processed to a personal dictionary for automatic translation (a part of the personal dictionary 132) when the determination in step 464 is negative. If the determination at step 464 is affirmative, nothing is done for that dictionary entry.

  The control of the routine executed when the speech translation result and the dictionary entry are received from the server 106 in the speech translation client 300 for using the speech translation service of the server 106 executed on the portable terminal 104. Structure.

<Operation>
"Overview"
─ Speech translation ─
A speech translation client 300 as shown in FIG. 3 is distributed in advance to the portable terminal 104 or the like. In this embodiment, it is assumed that the server 106 to which the portable terminal 104 can be connected is fixed by the speech translation client 300. Of course, if there are a plurality of servers 106, the user may select a desired one of them.

  The user operation when attempting to use the speech translation service of the server 106 is as described above. When the voice input by the user is completed, the recorded voice data is transmitted to the server 106 as a voice translation request together with information indicating the combination of translation languages, the terminal ID and address of the portable terminal 104. Prior to this, the user needs to call a setting screen in a predetermined procedure and select a combination of a source language and a target language to be used by the user.

  When the server 106 receives this speech translation request (step 402 in FIG. 5), it determines whether or not this session is new (step 404), and generates a unique session ID for the session when it is a new session. The session ID and the terminal ID of the partner terminal are recorded.

  Subsequently, at step 408, voice recognition is performed. At this time, the dictionary entry of the word dictionary for speech recognition in the source language is used. In step 410 following step 408, the voice data received from the portable terminal 104, the voice recognition result, and information indicating which dictionary entry is used are associated with the terminal ID and stored in the personal dictionary creation storage unit 380 (FIG. 3) is temporarily recorded.

  In the following step 412, a language pair is selected according to the language pair information in the speech translation request, and automatic translation is performed using the automatic translation word dictionary corresponding to the language pair and the target language word dictionary. In step 414, information indicating the dictionary entry used in the automatic translation process among the morpheme string of the automatic translation result, the automatic translation dictionary, and the word dictionary of the target language is temporarily stored in the personal dictionary creation storage unit 380. Recorded.

  Further, in step 416, reverse translation for the translation result and speech synthesis for the translation result are performed.

  Finally, in step 418, the speech recognition result, the translation result, the reverse translation thereof, and the synthesized speech of the translation result are transmitted to the portable terminal 104. At this time, in accordance with the information temporarily recorded in the personal dictionary creation storage unit 380 in Step 410 and Step 414, each dictionary entry used in the speech translation process is displayed together with information indicating the language and type of the dictionary. It is transmitted to the portable terminal 104.

  When the processing of step 418 is completed, the processing of the speech translation request from the portable terminal 104 is temporarily terminated, and the server 106 returns to a state of waiting for the next request (step 402).

--Adding an entry to the personal dictionary--
The portable terminal 104 that has received the information transmitted from the server 106 in step 418 operates as follows. Referring to FIG. 6, in steps 450, 452, and 454, the speech recognition result, the translation result, and the reverse translation are displayed, and the synthesized speech of the translation result is uttered. Further, by executing step 456 in FIG. 6, the speech recognition word dictionary (user dictionary 130, personal dictionary 132, and basic dictionary 134) in the local dictionary 110 among the speech recognition dictionary entries received from the server 106. If there is an entry not in the list, add the entry. Nothing is done for entries already in the speech recognition word dictionary. This process is performed for both the source language word dictionary and the target language word dictionary. In step 462, the same processing is performed for the automatic translation dictionary corresponding to the automatic translation language pair in the local dictionary 110. Thereafter, the portable terminal 104 returns to a state waiting for an operation by the user.

  When the speech translation service of the server 106 is used when the portable terminal 104 can communicate with the server 106 by the above processing, among the entries of the word dictionary and translation dictionary used at that time, the local dictionary of the portable terminal 104 Dictionary entries that do not exist in 110 are automatically added to the appropriate personal dictionary 132 in the local dictionary 110 of the portable terminal 104. The phrase used by the user is likely to be used repeatedly thereafter. When the local speech translation service 112 is operated on the portable terminal 104, the personal dictionary 132 is used, so that the possibility that the correct word is used in speech recognition and speech translation can be increased. As a result, when the portable terminal 104 cannot communicate with the server 106, it is possible to improve the accuracy of the speech translation service that is locally executed by the portable terminal 104.

  In the embodiment described above, entries used for the word dictionary of the source language for speech translation, the translation dictionary corresponding to the language pair, and the word dictionary of the target language are automatically added to the personal dictionary 132. However, the present invention is not limited to such an embodiment. Entries may be added only to some of the dictionaries instead of all the three types of dictionaries described above. Further, prior to adding the dictionary entry transmitted from the server 106 to the personal dictionary 132, the user may be allowed to confirm the contents, or the words added to the personal dictionary 132 and the words not to be added may be classified. .

[Second Embodiment]
<Outline>
In the first embodiment, when the portable terminal 104 and the server 106 are communicable and the user of the portable terminal 104 uses the speech translation service of the server 106, it is used for processing in the server 106. The registered dictionary entry is registered in the local personal dictionary of the portable terminal 104. In this way, once used vocabulary can be used locally on the portable terminal 104. When the portable terminal 104 cannot communicate with the server 106, even when the local speech translation service of the portable terminal 104 is used, the vocabulary can be used to improve processing accuracy.

  However, according to such a method, it is not registered in the personal dictionary unless it is a vocabulary that has been used once in connection with the processing of the speech translation request by the user. When the user frequently uses the speech translation service of the server 106 and the vocabulary required for the processing is wide, it is predicted that the personal dictionary of the portable terminal 104 will be enriched. However, in practice, it is difficult to use the speech translation service as frequently as the vocabulary of the personal dictionary increases. Therefore, it is more desirable to have a technique for quickly expanding a dictionary prepared locally in the portable terminal 104 by some method even if the frequency of use by the user is low. The speech translation system according to the second embodiment described below is such a system.

"overall structure"
Referring to FIG. 7, a speech translation system 500 according to the second embodiment includes a speech translation server 506 that provides an online speech translation service via the Internet, and communicates with the speech translation server 506 via the Internet. A plurality of portable terminals 504, 508, and 510 using the speech translation service by the speech translation system 500. The portable terminals 504, 508, and 510 do not need to have the same hardware and the same software configuration, but in any case, the voice of the first embodiment for using the voice translation service by the voice translation server 506 is used. A speech translation client similar to the translation client 300 is installed. Hereinafter, the communication sequence with the speech translation server 506 will be described by taking the portable terminal 504 as an example, on behalf of these portable terminals.

  When the portable terminal 504 can communicate with the speech translation server 506, it can use the speech translation service provided by the speech translation server 506 and cannot communicate with the speech translation server 506, as in the first embodiment. First, a local speech translation service is executed using resources locally provided in the portable terminal 504. For this purpose, the portable terminal 504 is provided with a local dictionary 520. The local dictionary 520 includes a dictionary for speech recognition, a dictionary for automatic translation, and a dictionary for speech synthesis. In the following description, in the present embodiment, for the sake of brevity, a dictionary for speech recognition, a dictionary for automatic translation, and a dictionary for speech synthesis are not particularly distinguished, and are simply referred to as a local dictionary.

  As shown in the upper left of FIG. 7, the local dictionary 520 initially includes a basic dictionary 522 that includes basic vocabulary necessary for speech translation, and a user dictionary 524 in which a user can register an entry. However, by executing the processing according to the present embodiment, the local dictionary 520 includes both the basic dictionary 522 and the user dictionary 524 as well as the user dictionary 524 and the basic dictionary 522 as illustrated in the lower left of FIG. A user supplement dictionary 554 that includes words that are not registered and that is likely to be used by the user is included. This user supplement dictionary 554 includes a large number of user dictionaries such as the portable terminals 508 and 510 including the portable terminal 504, and data mining using the vocabulary recorded in the user dictionary 524 of the portable terminal 504 as a clue. Using this method, a vocabulary that is likely to be used in the portable terminal 504 in the future is extracted from the vocabulary of the accumulated user dictionary and created. The data mining technique is a technique often used in mail order sales via the recent Internet. For example, it is used when comparing a purchase history of a certain user with a purchase history of another user and recommending to the user an article that is likely to be further purchased by the user.

  There are various data mining techniques. In the present embodiment, the frequent pattern mining method is used among them. Hereinafter, the overall data flow will be described with reference to FIG. 7, and then the software configuration for creating the user supplement dictionary in the speech translation server 506 will be described with reference to FIG.

  Referring to FIG. 7, portable terminal 504 transmits user dictionary 524 to speech translation server 506 at a certain timing (step 530). This transmission may be performed, for example, by calling a certain process when the user wants to obtain or update the user supplement dictionary, or may be performed at a certain cycle (for example, once a month). . The portable terminal 504 does not spontaneously send the user dictionary 524 to the speech translation server 506, but the user dictionary 524 is sent from the speech translation server 506 side to the portable terminal 504 that has used the speech translation server 506 in the past. You may request to send regularly.

  Similar processing is performed not only for the portable terminal 504 but also for the portable terminals 508 and 510. In this way, the user dictionary of a large number of portable terminals is always accumulated in the speech translation server 506 (step 532). When a user dictionary is transmitted again from a certain portable terminal, the old user dictionary from the portable terminal is replaced with a new dictionary.

  Note that the timing for collecting user dictionaries from a portable terminal is usually different from each other. It is not necessary to transmit the user dictionary to the speech translation server 506 from all portable terminals at the same time.

  If the user dictionary accumulated in this way is held, the speech translation server 506 analyzes the accumulated data periodically (for example, once a day) using the frequent pattern mining method (step 534). A software configuration for realizing this method will be described later. This frequent pattern mining method extracts a list of frequently used pairs of a word and a word registered in the same user dictionary as that word (that is, a word pair co-occurring in the same dictionary). Is done. The extracted word pair is called a frequent word pair.

  Furthermore, using the analysis result obtained by using the frequent pattern mining method, the speech translation server 506 identifies a supplement word to be registered in the user supplement dictionary 554 for each portable terminal (step 536). Specifically, the speech translation server 506 searches for frequent word pairs extracted by the frequent pattern mining method in which only one word is registered in the user dictionary. Then, the other word of the frequent word pair is selected as a supplement word for the dictionary of the portable terminal. This set of supplemental words is stored together with the terminal ID of the portable terminal that has transmitted the target user dictionary.

  Next, the portable terminal 504 sends a user supplement dictionary transmission request to the speech translation server 506 at some timing (step 550). This transmission request may be based on an instruction from the user, for example, or may be automatically transmitted after a predetermined time (for example, one day) has elapsed since the user dictionary was transmitted to the speech translation server 506 in Step 530. Also good.

  Upon receiving this transmission request, the speech translation server 506 reads a supplement word group corresponding to the terminal ID of the portable terminal 504 from the storage device and transmits it to the portable terminal 504 (step 538).

  The portable terminal 504 that has received the supplement word group generates a user supplement dictionary including the supplement word group in the local dictionary 520 (step 552). If the old user supplement dictionary is already in the local dictionary 520, the old dictionary is rewritten with the new user supplement dictionary.

<Software configuration>
The process executed by the portable terminal 504 can be easily realized by a so-called person skilled in the art. Therefore, the control structure of the program that realizes steps 534 and 536 in FIG. 7 among the processes executed in the speech translation server 506 will be described here.

"server"
In the following description, it is assumed that a large number of user dictionaries are already accumulated in the speech translation server 506. Referring to FIG. 8, the program for extracting words for the user supplement dictionary executed by the speech translation server 506 is a word pair (frequent word pair) that frequently appears in the accumulated user dictionary by the frequent pattern mining method. ) And a step 536 for specifying a set of words for the supplementary dictionary for each mobile terminal from the frequent word pairs extracted in step 534.

  Step 534 extracts all combinations of vocabulary pairs for all user dictionaries, and examines their appearance frequency (frequency at which the words constituting the vocabulary pairs are present in the same dictionary). And deleting a vocabulary pair whose appearance frequency is less than a predetermined threshold value among the vocabulary pairs. The threshold here varies depending on the number of words included in all user dictionaries, and is difficult to determine uniformly. In this case, rather than determining a threshold value and then finding an occurrence frequency word pair that satisfies the threshold value, the maximum number of word pairs to be extracted can be stored in the portable terminal 504. The so-called “top-K frequent set discovery” method is used in which the threshold is determined so as to obtain such a number of word pairs. By using this method, a word pair can be identified relatively efficiently, although the processing time is somewhat longer. For the top-K frequent set discovery method, see Reference 1 described below.

[Reference 1]
Tomoaki Uno, Hiroki Arimura, Introduction to Frequent Pattern Discovery Algorithm —From Item Set to Graph—, Proceedings of National Conference of the Japanese Society for Artificial Intelligence (CD-ROM), Vol.

  Step 536 includes step 580 of repeating the following processing for each user dictionary stored in the speech translation server 506.

  In step 580, for each of the vocabulary pairs extracted in step 534 and in step 582 for clearing the work area for storing supplemental words, only one vocabulary of the vocabulary pair is registered in the user dictionary. Step 584 for repeatedly storing the other vocabulary in the work area as a supplementary vocabulary for the user dictionary, and the vocabulary group saved in the work area in Step 584 are used as the user dictionary to be processed. And step 590 of storing in the storage device in association with the terminal ID of the terminal that has transmitted the message.

  Step 584 determines whether or not only one vocabulary of the processing target vocabulary pair is registered in the processing target user dictionary, and when the determination in step 586 is affirmative, And 588 adding the vocabulary to the work area. The determination in step 586 can be made by comparing each vocabulary constituting the vocabulary pair with the vocabulary in the user dictionary stored in the speech translation server 506.

<Operation>
Referring to FIG. 7, portable terminals 508, 510, etc., including portable terminal 504, each transmit a user dictionary to speech translation server 506 at each timing (step 530). The speech translation server 506 temporarily accumulates the user dictionary in the storage device every time it is transmitted (step 532). When the user dictionary is transmitted from the portable terminal that has already transmitted the user dictionary, the speech translation server 506 replaces the old dictionary with the new dictionary.

  For example, once a day, the speech translation server 506 processes the collected user dictionary data using a frequent pattern mining method, and determines a predetermined number of frequent vocabulary pairs (a predetermined number of appearance frequencies from the top). Are extracted (step 534). Specifically, steps 570 and 572 in FIG. 8 are executed.

  Subsequently, the speech translation server 506 executes the following processing for each of the accumulated user dictionaries. First, the work area is cleared (step 582 in FIG. 8). For each of the predetermined number of frequent vocabulary pairs extracted in step 534, if only one of the vocabularies constituting the vocabulary pair is registered in the user dictionary to be processed (YES in step 586), the vocabulary is stored in the work area. (Step 588). If both vocabularies constituting the vocabulary pair are not registered in the user dictionary to be processed or both are registered, nothing is done. By repeating such processing, vocabulary to be supplemented for the user dictionary to be processed is accumulated in the work area.

  When the processing for all frequent vocabulary pairs is completed, the speech translation server 506 stores the vocabulary stored in the work area in the storage device in association with the terminal ID corresponding to the user ID of the work target (step 590).

  The speech translation server 506 executes the processes in steps 580, 582, 584 and 590 described above for all user dictionaries. Therefore, when step 538 is completed, a set of vocabularies to be supplemented to the user dictionary is stored in the storage device of the speech translation server 506 in association with the terminal ID for all portable terminals that have transmitted the user dictionary. Will be.

  Referring to FIG. 7 again, when portable terminal 504 requests a supplement dictionary at a certain timing (step 550), speech translation server 506 replenishes the supplement stored in the storage device in association with the terminal ID of portable terminal 504. The word group is transmitted to the portable terminal 504 (step 538). Receiving this, the portable terminal 504 creates a new user supplement dictionary 554 with the transmitted supplement word group and stores it in the local dictionary 520. If the user supplement dictionary 554 already exists in the local dictionary 520, the portable terminal 504 replaces the old user supplement dictionary 554 with a user supplement dictionary composed of new words.

<< Effects of Embodiment >>
According to the second embodiment, a large number of user dictionaries are accumulated by the speech translation server 506, and frequent word pairs are extracted therefrom. For each user dictionary, when only one vocabulary of frequent word pairs is registered, the other vocabulary is also registered in the supplementary dictionary. Otherwise, no vocabulary is registered in the supplementary dictionary. Therefore, the vocabulary that is highly likely to be registered in the dictionary together with the vocabulary registered in the user dictionary and is not yet registered in the user dictionary is registered in the user supplement dictionary. Even a vocabulary that the user has not used when using the speech translation service is registered in the user supplement dictionary, and the user dictionary can be efficiently reinforced. Furthermore, since frequent word pairs are extracted from the result of accumulating a large number of user dictionaries, it is possible to efficiently supplement words that may be used in the future without being limited to the vocabulary range used by the user. Furthermore, since the user supplementary vocabulary increases every time the user translation dictionary 506 is requested to update the user supplementary dictionary, the speed of reinforcing the user dictionary can be increased.

  Note that frequent pattern mining can be realized by using an a priori method, an FP-growth method, an association rule, and the like in addition to the above-described method. In short, in this embodiment, user dictionary data collected from a large number of terminals is used to determine the vocabulary to be supplemented to the user dictionary of a specific terminal. Any method that can find a word that is highly likely to be registered in the user dictionary together with the word registered in the dictionary may be used.

[Possible variants]
In the first embodiment, the vocabulary is automatically registered in the personal dictionary instead of the user dictionary. Also in the second embodiment, the vocabulary of a user supplement dictionary different from the user dictionary is extracted. However, the present invention is not limited to such an embodiment. Vocabulary may be added to the user dictionary itself instead of a dictionary different from the user dictionary. However, in this case, it is impossible to distinguish the words that the user has added consciously. Therefore, it is desirable to take measures to distinguish the words if possible. Further, the vocabulary reinforcement in the second embodiment may be applied to the vocabulary in the personal dictionary in the first embodiment. Furthermore, a dictionary in which vocabularies are automatically registered in the first embodiment is called a personal dictionary, and a dictionary in which vocabularies are automatically accumulated in the second embodiment is called a user supplement dictionary. However, these are merely dictionary names, and it goes without saying that other names may be used.

  In the second embodiment, a plurality of portable terminals use a single server to request creation of a user supplement dictionary. However, the present invention is not limited to such an embodiment. A plurality of servers may be provided, and each server may create a user supplement dictionary for different portable terminals. In this case, it is more preferable in terms of supplementing the vocabulary if the user dictionary data accumulated at each server is shared between the servers. However, if the number of accumulated user dictionaries is too large or the vocabulary registered in each user dictionary is too large, the processing time on the server is affected. Therefore, the number of portable terminals processed by each server may be limited.

  In the second embodiment, frequent word pairs are further extracted from the accumulated user dictionary data by data mining. However, the present invention is not limited to such an embodiment. For example, instead of frequent word pairs, one that frequently appears in a set of three words or one that frequently appears in a group of four or more words may be extracted.

  In the description of the above embodiment, the language to be used is not particularly limited. However, as described above, it is needless to say that a combination of languages for translation can be designated. In that case, the above-mentioned dictionary is provided for each language for the word dictionary, and it is necessary to provide the translation dictionary according to the combination of each language and the translation direction, and the above-described processing needs to be implemented for each. is there.

  In the above embodiment, the speech translation service is taken as an example. However, the present invention is not limited to such services. For example, the present invention can be applied to a service that provides online input by voice recognition alone, an online translation service based on text input, and the like. That is, the language service to which the present invention is applicable is not limited to the one that uses voice. It is a language service that can be used by switching between a service provided by a server online and the same type of application that is executed locally on the terminal, and can be applied to any service that uses a so-called dictionary. .

  In the above-described embodiment, when the client device makes a local dictionary reinforcement request in some form, the server creates local dictionary reinforcement information and transmits it to the client device. However, the present invention is not limited to such an embodiment, as long as information for reinforcing language resources used locally is created using the results of services provided by the server. The present invention can also be applied to a system in which a server pushes and distributes to a client device without a request from the client device.

  The embodiment disclosed herein is merely an example, and the present invention is not limited to the above-described embodiment. The scope of the present invention is indicated by each claim of the claims after taking into account the description of the detailed description of the invention, and all modifications within the meaning and scope equivalent to the wording described therein are included. Including.

100, 500 Speech translation system 104, 504, 508, 510 Portable terminal 106, 506 Server 108 Large vocabulary dictionary 110, 520 Local dictionary 130, 524 User dictionary 132 Personal dictionary 134, 522 Basic dictionary 250 Processor 252 Memory 356 CPU
360 RAM
380 Personal dictionary creation storage unit 554 User supplement dictionary

Claims (6)

A service using means for using the language service by communicating online with a service providing facility device that provides a language service using a language resource prepared in advance;
Local service execution means for executing a service of the same type as the language service locally using a local language resource having a smaller scale than the language resource when communication with the service providing equipment is not possible;
A request transmitting means for transmitting a request for receiving provision of the service to the service providing equipment device in response to an input to the language service when communication with the service providing equipment device is possible;
When communication with the service providing equipment is possible, a language resource for reinforcing the local language resource is added to the execution result of the language service transmitted from the service providing equipment in response to the request. And a language resource reinforcement means for augmenting the local language resource using the received language resource in response to the received language resource. In response to the request from the request transmission unit, the service providing equipment is used in the process of obtaining the processing result of the language data to be processed by the language service and the processing result of the language resource. A language resource for reinforcement consisting of language resources is transmitted together or separately to the local language resource reinforcement device;
The language resource reinforcement means receives the reinforcement language resource transmitted from the service provision facility device when it can communicate with the service provision facility device, and sends the reinforcement language resource to the local language resource. The local language resource reinforcement device according to claim 1, further comprising language resource addition means for adding a resource. The local language resource reinforcement device further transmits the local language resource to the service providing equipment device at a designated timing independent of the transmission of the request when communication with the service providing equipment device is possible. The local language resource reinforcement device according to claim 1, further comprising a local language resource transmission unit for performing the operation. Service providing means for providing a language service using a language resource prepared in advance to other terminal devices via online communication;
Reinforcing language resource creating means for creating a reinforcing language resource for reinforcing a local language resource used locally in another terminal device by using a service execution result by the service providing means;
Reinforcing language resource transmitting means for transmitting the reinforcing language resource created by the reinforcing language resource creating means to the other terminal device when communicating with the other terminal device; A service providing equipment device,
The local language resource includes a plurality of data items that are distinguished from each other,
The reinforcing language resource creating means includes:
Local language resource storage means for collecting and storing local language resources used in each terminal device from a plurality of terminal devices;
Frequent data item group specifying means for specifying one or more frequent data item groups based on the local language resources stored in the local language resource storage means, and the one or more frequent data item groups Each comprises a plurality of data items that are more likely to co-occur within each of said local language resources than a threshold;
The reinforcing language resource creating means further includes:
For each of the plurality of terminal devices, identify a frequent data item group in which only some data items are included in the local language resource transmitted from the terminal device in the one or frequent data item group, A service providing equipment device including means for creating the reinforcing language resource for the terminal device based on the frequent data item group.   In response to receiving a request from the other terminal device for requesting reinforcement of a local language resource used locally in another terminal device, the service providing unit creates a service. The service providing facility apparatus according to claim 4, further comprising means for creating the reinforcing language resource using the partial language resource used when providing the service. The means for creating includes a partial language resource used when the service providing means provides a service in response to a request from the other terminal device among the previously prepared language resources. Including temporary storage means for temporary storage;
6. The service provision according to claim 5, wherein the reinforcement language resource transmitting unit transmits the partial language resource stored in the temporary storage unit together with a processing result of the service provided by the service providing unit. Equipment equipment.

Images