JP2006018336A - Meta data generation device and method, and meta data generation program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To extract meta data including attribute information from a message inputted by natural language processing. <P>SOLUTION: This meta data generation device is constituted of a message input part 101, a message analysis part 102, a unique expression selecting part 103, a meta data registering part 104, a meta data storing part 105 and a control part 106. The unique expression and the attributes are estimated from text data inputted as a message by a user, and the unique expression having the attributes related with media data and the attributes are registered as meta data, so that it is possible to simplify meta data application to the media data. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a metadata generation apparatus that generates metadata of media data.

  In recent years, attempts have been made to add metadata to media data to facilitate access to media data, particularly stream media data (programs sent on television broadcasts, movies provided on DVDs, etc.), for example, MPEG. -7 etc.

Metadata refers to data describing information about certain data.
For example, in television broadcasting, by describing the contents of a program related to the media data and information on the performers in the program in media data such as video and audio, the desired content and media data relating to the performers are stored. This can be useful when searching. Alternatively, it is possible to respond to a request to selectively search and view only a specific part in the media data, for example, only a scene where a B player of Team A appears in a baseball program. That is, the scene can be easily retrieved by preliminarily describing the data indicating the scene in which the B player appears in the program as metadata along with the media data.

  As a metadata notation method for multimedia contents, MPEG-7 established in the MPEG (Moving Picture Experts Group) which is an international standard is known.

  Conventionally, as means for adding metadata to media data, conventionally, (1) pattern information such as voice and telop in media data is converted into text information by media conversion such as voice recognition and OCR (for example, patents). There are known methods for inputting metadata in association with media data manually using a dedicated tool on a computer called an authoring tool (see Reference 1) and (2).

JP 2003-296333 A JP 2003-178076 A

  Conventional methods for adding metadata to media data are as follows: (1) In the method of extracting text information as metadata from media data by media conversion such as voice recognition or OCR, the limit of media conversion capability (recognition) Accuracy, etc.), there is a problem that it can only be applied to limited types of media data (for example, news programs), and text information is directly input as metadata, so what is the metadata (for example, a baseball team) There is no attribute information as to whether it is a baseball player or the like.

  (2) In the manual method using a dedicated tool, the metadata must be input in a fixed format one by one, and the burden on the operator is great, and the quality and quantity of the metadata to be assigned are There was a problem of depending on the knowledge of the person.

SUMMARY OF THE INVENTION In view of the above-described conventional problems, an object of the present invention is to extract metadata including attribute information from a message input by natural language processing.
Moreover, it aims at providing the metadata production | generation apparatus which can produce | generate the metadata which does not depend on the knowledge of one user by analyzing the message which several users exchange.

  In order to achieve the above object, according to the present invention, in a metadata generation apparatus for generating metadata of media data, message input means for inputting message information related to the media data, and input by the message input means A message analysis unit that extracts a phrase and an attribute of the phrase from the extracted message information; a phrase selection unit that selects a phrase having an attribute related to the media data from an extraction result of the message analysis unit; There is provided a metadata generation apparatus comprising: a metadata registration unit that stores the phrase and the attribute of the phrase in the storage unit as the metadata.

In order to achieve the above object, in the present invention, in a metadata generation method for generating metadata of media data,
A message input step for inputting message information related to the media data, a message analysis step for extracting words and attributes of the words from the message information input by the message input step, and an extraction result of the message analysis step. A phrase selecting step for selecting a phrase having an attribute related to the media data from the media, and a metadata registration step for storing the selected phrase and the attribute of the phrase in the storage means as the metadata. A metadata generation method is provided.

  In order to achieve the above object, according to the present invention, in a metadata generation program for causing a computer to generate metadata of media data, the computer is caused to input message information related to the media data, and the input A phrase and an attribute of the phrase are extracted from the extracted message information, a phrase having an attribute related to the media data is selected from the extraction result, and the selected phrase and the attribute of the phrase are used as the metadata. Provided is a metadata generation program that is stored in a storage means.

The present invention relating to the apparatus is also established as an invention relating to a method, and the present invention relating to a method is also established as an invention relating to an apparatus.
Further, the present invention relating to an apparatus or a method has a function for causing a computer to execute a procedure corresponding to the invention (or for causing a computer to function as a means corresponding to the invention, or for a computer to have a function corresponding to the invention. It can also be realized as a program (for realizing the program), and can also be realized as a computer-readable recording medium on which the program is recorded.

  As described above, according to the present invention, metadata including attribute information can be easily extracted from a message input by natural language processing.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
(First embodiment)
FIG. 1 is a diagram showing a configuration of a metadata generation apparatus according to the first embodiment of the present invention.
In FIG. 1, the metadata generation apparatus includes a message input unit 101, a message analysis unit 102, a specific expression selection unit 103, a metadata registration unit 104, a metadata storage unit 105, and a control unit 106. This metadata generation apparatus can be realized, for example, on a general computer (computer) shown in FIG.

FIG. 2 is a diagram illustrating an example of hardware when the metadata generation apparatus of the present invention is realized by a computer.
The computer includes a central processing unit 201 that executes a program, a memory 202 that stores the program and data being processed by the program, and a magnetic disk drive 203 that stores the program, data to be searched, and an OS (Operating System). And an optical disc drive 204 for reading and writing programs and data on the optical disc.

  Furthermore, an image output unit 205 that is an interface for displaying a screen on a display, an input receiving unit 206 that receives input from a keyboard, mouse, touch panel, and the like, and an input / output interface (for example, USB (Universal Serial) Bus), an audio output terminal, etc.). Further, the display device 208 includes a display device 208 such as an LCD, a CRT, and a projector, an input device 209 such as a keyboard and a mouse, and an external device 210 such as a memory card reader and a speaker. The external device 210 may be a network instead of a device.

  The central processing unit 201 implements each function shown in FIG. 1 by executing a program after reading the program from the magnetic disk drive 203 or the optical disk drive 204 and storing it in the memory 202. During execution of the program, part or all of the search target data may be read from the magnetic disk drive 203 and stored in the memory 202.

  The basic operation is to receive a search request from a user via the input device 209 and search for search target data stored in the magnetic disk drive 203 or the memory 202 in response to the search request. Then, the search result is displayed on the display device 208.

  The search result may be displayed not only on the display device 208 but also presented to the user by voice by connecting a speaker as the external device 210, for example. Alternatively, a printer may be connected as the external device 210 and presented as a printed matter.

  The message input unit 101 in FIG. 1 can be realized by inputting text data using the keyboard of the input device 209 in FIG. In addition to a keyboard or the like, there may be provided conversion means for recognizing the character pattern in the audio information or image information included in the media data and converting it into text information, and the output of this conversion means may be input as a message. .

  FIG. 3 shows an example of a graphical user interface (GUI) screen displayed on the display device 208 of FIG. 2 when inputting this text data. This GUI screen has a video display area 301, a video control area 302, a video file name input area 303, a message input area 304, and a message control area 305.

  The video display area 301 is for displaying a reproduction of video data stored in, for example, the magnetic disk drive 203 of FIG. When the user operates the “play”, “stop”, “rewind”, and “fast forward” buttons arranged in the video control area 302 using the mouse or keyboard of the input device 209, the video data is played back. Can be controlled.

  In this embodiment, the media data is described as an example of a television broadcast program or a DVD that includes an audio signal together with a video signal, but the media data is not limited to this. For example, in the case of a program by radio broadcasting, the media data does not include a video signal but only an audio signal.

  In this specification, media data refers to data representing media to be viewed by humans, combining text, numerical data, images, audio, video, and the like.

  When starting the metadata generation process, the user first inputs the file name of the video data to be viewed in the video file name input area 303, and clicks the “START” button in the message control area 305 to input the input device 209 in FIG. 2. Press with the mouse. As a result, processing for generating metadata regarding the media data of the input file name is started.

  Next, when a “play” button in the video control area 302 is pressed, video data having a file name specified in the video file name input area 303 is displayed in the video display area 301.

  Thereafter, the user inputs text data as a message through the message input area 304 using the keyboard in the input device 209 of FIG. 2 while viewing the video displayed in the video display area 301. When the input of one message is completed, the user presses the “Send” button in the message control area 305 in FIG. 3 with the mouse. Then, one message input to the message input area 304 is transmitted to the message input unit 101 in FIG. 1, and the message input to the message input area 304 is cleared. If the user wants to cancel the message input in the message input area 304, the user presses the “Clear” button in the message control area 305.

When the message input for one piece of video data is completed, the user presses an “end” button in the message control area 305.
In the present invention, the video display area 301 and the video control area 302 are not essential.
Further, the video data may be obtained from the outside of the metadata generation apparatus through the external apparatus / network 210 in FIG. 2 instead of the magnetic disk drive 203 in FIG.

  The message analysis unit 102, the specific expression selection unit 103, the metadata registration unit 104, and the control unit 106 in FIG. 1 are realized as software executed by the CPU 201 and the memory 202 in FIG. 2, respectively, and the processing contents will be described later. The metadata storage unit 105 is realized by, for example, the magnetic disk drive 203 in FIG.

  In this embodiment, the metadata accumulation unit 105 that accumulates metadata is not necessarily in the metadata generation apparatus. For example, you may accumulate | store in the server connected with the network.

The control unit 106 calls the message analysis unit 102, the specific expression selection unit 103, and the metadata registration unit 104 in order, and exchanges data among them.
When the “start” button in the message control area 305 in FIG. 3 is pressed, the control unit 106 stores the video file name input in the video file name input area 303 in the internal memory 202 or the like, and Then, the metafile corresponding to the video file name is initialized.

  The control unit 106 uses the created video file name of the metadata as a file name obtained by replacing the extension part of the video file name, for example. For example, when the video file name of the media data is “sample.av” (av is the extension of the video file), “sample.md” (md is the extension of the metadata file) as the metadata file name. ). In this way, the part excluding the extension of the file name, in this example, “sample” is made the same so that the control unit 106 can handle this video file if the video file name is known when using the metadata. You can easily find a metafile to do.

The control unit 106 also controls playback control (playback, stop, rewind, fast forward) of video data on the video display area 301.
FIG. 4 is a flowchart showing the flow of message analysis processing in the message analysis unit 102.
First, in step S401, the message analysis unit 102 extracts a morpheme and a part of speech and an attribute for each morpheme from text data that is an input message based on the morpheme analysis dictionary 451.

  The morpheme analysis dictionary 451 stores part-of-speech for each morpheme and attributes necessary for the subsequent proper name extraction process S402. As this attribute, for example, the attribute “animal” is retained in the morpheme “dog” and the attribute “plant” is retained in the morpheme “azalea”.

Here, the attribute is the following information.
・ Information on semantic classification of nouns (above “dog” is [animal], azalea is [plant], etc.)
・ Information about words that co-occur with proper expressions ("President" is [Personal co-occurrence word], "Ku" is [Geo-co-occurrence word], "Gram" is [Weight unit], etc.],
・ Information about the proper expression itself (“London” is [Capital Name], “Guardian Gate” is [Building], etc.).
For each morpheme extracted from the message, its part of speech and attribute are also passed from the morpheme analysis process S401 to the proper name extraction process S402. Note that the morphological analysis processing in step S401 is a known technique used in natural language processing such as machine translation and is not the gist of the present invention, and thus the description thereof is omitted.

  Next, in step S402, the message analysis unit 102 extracts a list of a specific expression and a set of the attribute from the morpheme passed through the process of step S401 and its part of speech and attribute based on the specific name extraction rule 452, and the specific expression The message is passed to the selection unit 103, and the message analysis process ends. Specifically, first, a morpheme of a noun is set as a single specific expression, and an attribute of the morpheme is set as an attribute of the specific expression. Next, it is tried based on the proper name extraction rule 452 whether each proper expression can be extended back and forth. The proper name extraction rule 452 describes a pattern that can be expanded and an attribute when the morpheme is expanded by a combination of adjacent morpheme attributes before and after the morpheme that has become a single unique expression. For example, if morphemes having “place name” and “address” are adjacent to each other, the connected one is set as a new unique expression, and the attribute of the specific expression is set as “address”. When there is no applicable rule in the unique name extraction rule 452, the message analysis unit 102 passes the list of combinations of unique expressions and their attributes to the unique expression selection unit 103 and ends the message analysis process.

  Regarding the processing of extracting the above-described specific expressions and their attributes, for example, “Ichimura, Saito, Sakai, Kokubun, Koyama:“ Study on the relationship between question answering, Japanese specific expression extraction and proper expression system ”, The technology known by the Information Processing Society of Japan, Natural Language Processing Study Group, NL-161-3, 2004. ”, etc. may be used.

  Next, the specific expression selection unit 103 selects only a specific expression (word / phrase) having a predetermined attribute from the list of combinations of the specific expression (word / phrase) and its attributes passed from the message analysis unit 102. Then, a list of combinations of the selected unique expressions (phrases) and their attributes is passed to the metadata registration unit 104. The predetermined attribute is, for example, an attribute related to video data such as a person name or a song title, and the specific expression selecting unit 103 selects only a specific expression (word) having this attribute.

Next, the control unit 106 passes the metadata file name to the metadata registration unit 104.
Next, the metadata registration unit 104 adds, to the metadata file name data received from the control unit 106, a list of combinations of the specific expressions selected by the specific expression selection unit 103 and their attributes as metadata. At this time, the metadata registration unit 104 reads metadata already registered in the metadata file in advance, and adds a list of combinations of unregistered unique expressions and their attributes as metadata.

(Modification)
As a modification of the present embodiment, the following may be performed.
Furthermore, section detecting means is provided for detecting a section in which the feature amount falls within a predetermined range from the media data based on a change in the feature amount of the image feature amount or audio feature amount of the media data. For example, this section detection means is provided to the control unit 106 as a function. Then, holding means for holding the correspondence relationship between the message input to the message input unit 101 and the position on the time axis of the media data based on the input time of the message is provided. As this holding means, a part of the magnetic disk drive 203 may be used, and the correspondence relationship may be held by the control unit 106. When registering the metadata, the metadata is registered in association with the section detected by the section detecting means based on the input time of the metadata extraction source message held by the holding means.
Thereby, metadata can be given as for a specific section, for example, a scene.

  As described above, according to the present embodiment, metadata including attribute information can be easily extracted from a message input by natural language processing. Since the created metadata can be displayed using attributes such as a personal name list and a song name list, the contents of the video data can be examined from an interesting viewpoint.

(Second Embodiment)
The second embodiment of the present invention is realized by a plurality of computers connected via a network 500 shown in FIG.
FIG. 5 shows a configuration in which a plurality of client computers 502 are connected to one server computer 501. Both the server computer 501 and the client computer 502 have the configuration shown in FIG.

  FIG. 6 shows a configuration on the client computer 502 side. In addition, the same number is attached | subjected about the same thing as FIG. 1 of 1st Embodiment, and description is abbreviate | omitted. The client computer 502 is newly provided with a communication unit 607 as compared with FIG.

The server computer 501 distributes video data determined by time to all the client computers 502.
The plurality of client computers 502 simultaneously receive and display the video data distributed by the server computer 501 via the network 500.
The screen configuration on the display device 208 in the client computer 502 is shown in FIG.
In FIG. 7, the screen includes a video data display area 701, a video data control area 702, a video file name display area 703, a message input area 704, a message control area 705, and a message display area 706.

The video data display area 701 displays video data received from the server computer 501 in real time.
On the client computer 502 side, reception of video data is refused until the “Receive” button in the video data control area 702 is pressed.
When the user presses the “Receive” button in the video data control area 702, the client computer 502 makes an inquiry to the server computer 501 via the communication unit 607 for the file name of the video data distributed from the server computer 501 at this time. The file name of the video data obtained as a result is displayed in the video file name display area 703 and stored in the internal memory 202 or the like by the control unit 607 in the client computer 502.

  When the “Receive” button is pressed, the client computer 502 informs the server computer 501 via the network 500 and starts receiving video data.

  The server computer 501 holds a list of client computers 502 that are receiving video data, for example, in the memory 202 in the server computer 501, and distributes video data to the client computer 502 that is viewing based on this list. To do. The server computer 501 updates the list when it is newly notified from the client computer 502 that the “receive” button has been pressed.

  When the client computer 502 depresses the “stop” button in the video data control area 702 or when the distribution of the video data from the server computer 501 is stopped, the display of the video data in the video data display area 701 is terminated, and Then, the metadata generation process for this video data is terminated.

  The user can input a message by inputting text information to the message input area 704 with the keyboard in the input device 209 while viewing the video data displayed in the video data display area 701.

  When the user presses the “Send” button in the message control area 705 after inputting a message in the message input area 704, the control unit 106 displays the message input in the message input area 704 in the message display area 706 and the message input part. 101 and the server computer 501 via the communication unit 607. This message is transmitted by the server computer 501 to the other client computer 502 receiving the video data when the message is received. If the user presses the “Clear” button after inputting a message in the message input area 704, the message input is canceled.

  When the client computer 502 receives a message from another client computer 502, the client computer 502 displays the received message in the message display area 706 and passes it to the message input unit 101.

  A message input from the keyboard in the input device 209 or a message received from another client computer 502 via the communication unit 607 is transmitted to the control unit 106 through the message input unit 101. The control unit 106 stores the transmitted message in the magnetic disk drive 204, for example.

  When the user presses the analysis button in the message control area 705 in FIG. 7, the control unit 106 reads out the message group accumulated so far in the magnetic disk drive 204 and passes it to the message analysis unit 102.

The message analysis unit 102 executes the process shown in FIG. 8 for each message sentence in the passed message group.
FIG. 8 is a flowchart for explaining the operation of the message analysis process.
In FIG. 8, first, in step S801, the sentence pattern determination process S801 determines whether each sentence in the message is a question sentence based on the sentence pattern 851. The sentence pattern 851 describes a sentence pattern of a question sentence using a regular expression. If one or more of these patterns are successfully matched, the sentence pattern is determined to be a question sentence, and neither of the patterns can be successfully matched. Is determined to be other than a question sentence.

  Next, morphological analysis processing S802 and specific expression extraction processing S803 are executed. The morpheme analysis process S802 corresponds to the morpheme analysis process S401 of FIG. 4, the specific expression extraction process S803 corresponds to the specific expression extraction process S402 of FIG. 4, and these processes are the message analysis of the first embodiment. Since the processing is the same as that of the unit 102, description thereof is omitted.

  After step S803, in step S804, if all the sentences in the message are determined to be other than the question sentence in the sentence pattern determination process S801, the message is directly passed to the specific expression selection unit 103 through the control unit 106. On the other hand, if it is determined that the sentence is a question sentence in one or more sentences in the message in the sentence type determination process S801, the answer type estimation process is performed on the message in the next step S805.

  In the answer type determination process in step S805, the result of the specific expression extraction process S803 and the sentence end expression of the sentence determined as the question sentence in the message are obtained as the answer to the question by the message based on the answer type estimation rule 854. Estimate the type of information you have.

  In the next step S806, a message next to the message including the question sentence is extracted and an answer extraction process is performed. After the answer extraction process, the message subject to the process is deleted from the accumulated magnetic disk drive 204 so that it is not subject to the processing after the sentence pattern determination process S801.

  Note that the “message” and “message sentence” are composed of a plurality of sentences, and each sentence of the plurality of sentences is described as a “message sentence”.

  Next to step S805, in the answer extraction process in step S806, the answer part is extracted by deleting a description specified in advance from the beginning of the sentence and the end of the sentence, and the extracted answer part is used as a unique expression. In addition, the type of information obtained as an answer, which is the result estimated in the answer type estimation process S805, is passed to the specific expression selection unit 103 as an attribute of this specific expression.

The specific expression selection unit 103, the metadata registration unit 104, and the metadata storage unit 105 perform the same processing as that in FIG. 1 described in the first embodiment.
As described above, according to the present invention, metadata including attribute information can be easily extracted from a message input by natural language processing. In the second embodiment, metadata can be created by a plurality of people sharing knowledge. In addition, it is possible to extract expressions and their attributes as metadata by analyzing a question sentence that has led to a message that could not be analyzed alone.

(Third embodiment)
As in the second embodiment, the third embodiment of the present invention is realized by a plurality of computers connected by a network shown in FIG. Also, the configuration implemented on the client computer 502 side is as shown in FIG. 6 and is not different from the second embodiment.

  However, in the second embodiment, the video data is not distributed simultaneously from the server computer 501 to the client computer 502, but the requested video data is distributed for each client computer 502 at the requested timing. Different.

  In addition, the server computer 501 includes a unit for storing a message for each video data, for example, in the magnetic disk drive 203. When the user of the client computer 502 views the video data, the other client computer that has viewed the same video data. Reference is made in accordance with the position on the time axis where the user inputs a message 502 and the user is viewing, and a reply to these messages can be input. For this reason, the message held by the server computer 501 stores the position on the time axis of the video data displayed when the message is input (for example, the relative time from the beginning of the video data) in association with it.

In addition, if there are different points, they will be described as needed.
The screen configuration on the display device 21 in the client computer 502 is the same as in FIG.
The video file name display area 703 of the third embodiment is the same as the video file name input area 303 of FIG. 3 described in the first embodiment. Each client computer 502 individually inputs a video file name of video data to be displayed in the video data display area 701 through the video file name display area 703, and presses a “receive” button in the video data control area 702, thereby The server computer 501 is requested to transmit video data with a file name. A keyboard or the like of the input device 209 is used for inputting a video file name of video data to be displayed.

  The server computer 501 distributes the video data of the video file name requested from the client computer 502 to the requested client computer 502, and at a predetermined time interval, the client computer 502 during each time interval. A message related to the video data transmitted from the server to the server computer 501, that is, a message transmitted to the server computer 501 in the past is transmitted. Then, the client computer 502 displays the video data received from the server computer 501 in the video data display area 701 and displays messages received from the server computer 501 at regular time intervals in the message display area 706.

  The user inputs a question message asking that he / she does not know the scene in which the video data is viewed while viewing the video data displayed in the video data display area 701 and the message displayed in the message display area 706. Or enter a message to answer a message from another user. A reply to another user's message, for example, an answer message to another user's question can be made by pressing the message. When the user presses a question message of another user in the message display area 706, the control unit 106 displays the message on the upper part of the message input area 704 in FIG. For example, a question message is displayed in the upper half of the message input area 704, and a space for the user to newly input an answer is created in the lower half. Thereby, the user can know that the client computer 502 is in a mode for inputting a reply to the displayed message.

  Reply messages input to the message input area 704 in this mode are stored in, for example, the magnetic disk drive 203 by the control unit 106 while holding the reply relation to the message by pressing the “Send” button in the message control area 705. In addition, the message display area 706 displays the reply relationship with the message (for example, “answer to the question”) in a format that is clearly indicated, and the input reply message is displayed in the input time (relative time from the top of the video data). ) And the message of the reply source are transmitted to the server computer 501.

  The server computer 501 adds the transmitted reply message to the message related to the video data together with the relationship between the input time (relative time from the top of the video data) and the message of the reply source. This reply message can be viewed according to the input time for the referring user.

  When inputting a new message such as a new question about video data, the user presses a “new” button in the message control area 705 and then inputs it to the message input area 704. After the input, when the user presses the “Send” button in the message control area 705, the new message is sent to the control unit 106 through the message input unit 101. The control unit 106 stores the new message and the input time of the new message (relative time from the beginning of the video data) in the magnetic disk drive 203 and simultaneously sends it to the server computer 501 via the communication unit 607. As a result, the server computer 501 is notified that a new message related to the video data has been input.

  To cancel a reply message or new message after it has been entered in the message input area 704, a “clear” button in the message control area 705 is pressed. When the “clear” button in the message control area 705 is pressed, the control unit 106 clears the display of the message input area 704 on the screen.

  When the “analysis” button in the message control area 705 is pressed, the message displayed in the message display area 706 is sent to the message analysis section 102 through the control section 106. The message analysis unit 102 extracts top layer messages one by one and performs the same processing as in the second embodiment. The second embodiment is different from the second embodiment in that regarding a message estimated as a question, when there is a message replied to this estimated message, the replied message is regarded as a response message, and a response extraction process is performed. S806 is in place.

In this embodiment, the metadata registration unit 104 stores the metadata separately at time intervals at which the server computer 501 switches messages.
Regarding the output of the message analysis unit 102 and the specific expression selection unit 103, first, it is checked whether the same metadata is stored in the time zone including the input time when the message from which the metadata is extracted is input. , And add it as metadata for that time zone.

As a modification, the metadata storage unit 105 may be provided on the server computer 501.
As described above, according to the present invention, metadata including attribute information can be easily extracted from a message input by natural language processing. Further, in the third embodiment, each user can exchange messages without viewing the same video data all at once, and the metadata extracted from the messages is not related to the entire video data but related parts. Can only be granted.

  Note that the present invention is not limited to the above-described embodiment as it is, and can be embodied by modifying the constituent elements without departing from the scope of the invention in the implementation stage. In addition, various inventions can be formed by appropriately combining a plurality of components disclosed in the embodiment. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, constituent elements over different embodiments may be appropriately combined.

The block diagram for demonstrating the structure of the metadata production | generation apparatus which concerns on 1st Embodiment. The figure which shows an example of the hardware when a metadata production | generation apparatus is implement | achieved with the computer. The figure which shows the example of the graphical user interface (GUI) screen displayed on the display apparatus 208 of FIG. 2 when inputting text data. 5 is a flowchart showing a flow of message analysis processing in the message analysis unit 102. The figure for demonstrating the structure with which the several client computer 502 was connected to the one server computer 501. FIG. The block diagram for demonstrating the structure by the side of the client computer 502. FIG. The figure for demonstrating the screen structure on the display apparatus 208 in the client computer 502. FIG. The flowchart for demonstrating operation | movement of the analysis process of a message.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 101 ... Message input part, 102 ... Message analysis part, 103 ... Specific expression selection part, 104 ... Metadata registration part, 105 ... Metadata storage part, 106 ... Control part, 201 ... Central processing unit, 202 ... Memory, 203 ... Magnetic disk drive, 204 ... Optical disk drive, 205 ... Image output unit, 206 ... Input reception unit, 207 ... I / O unit, 208 ... Display device, 209 ... Input device, 210 ... External device, 451, 852 ... Morphological analysis dictionary, 452 ... Morphological analysis rule, 607 ... Communication unit, 851 ... Character pattern, 853 ... Unique name extraction rule, 854 ... Answer type estimation rule.

Claims (8)

In a metadata generation device that generates metadata of media data,
Message input means for inputting message information related to the media data;
Message analysis means for extracting a phrase and attributes of the phrase from the message information input by the message input means;
A phrase selection unit that selects a phrase having an attribute related to the media data from the extraction result of the message analysis unit;
A metadata generation apparatus comprising: a metadata registration unit that accumulates the selected word and the attribute of the word in the storage unit as the metadata. Communication means for exchanging message information input by the message input means with another metadata generation device;
2. The metadata generation apparatus according to claim 1, wherein the message analysis unit extracts a phrase and its attribute in message information received from the other metadata generation apparatus via the communication unit. The message information includes time position information related to the position on the time axis of the media data according to the time when the message is input in addition to the message,
The metadata generation apparatus according to claim 1, wherein the metadata registration unit registers the metadata in association with the time position information for a message from which metadata is extracted. The message input means inputs message information including inter-message relationship information indicating a relationship with another message,
The message analysis means includes
Means for analyzing the message and determining that it is a question message including a question sentence;
A reply message to the question message is regarded as an answer message to the question message using the inter-message relation information, and for the question message, an answer type required by the question text in the question message is determined, and the answer message 4. The metadata generation apparatus according to claim 1, further comprising means for extracting a word / phrase to be an answer to the question sentence from the attribute and setting an attribute of the word / phrase to the answer type. 5. Furthermore, a conversion means for recognizing a character pattern in the audio information or image information included in the media data and converting it into text information,
5. The metadata generation apparatus according to claim 1, wherein the message input unit inputs an output of the conversion unit as a message. 6. Furthermore, based on the change of the feature amount of the image feature amount or the audio feature amount of the media data, a section detecting means for detecting a section in which the feature amount is within a predetermined range from the media data,
The message input means includes
Holding means for holding a correspondence relationship between the input message and the position on the time axis of the media data based on the input time of the message;
The metadata registration means includes:
6. The metadata is registered in association with a section detected by the section detection unit based on an input time point of the metadata extraction source message held by the holding unit. The metadata generation apparatus of any one of Claims. In a metadata generation method for generating metadata of media data,
A message input step for inputting message information related to the media data;
A message analysis step of extracting a phrase and an attribute of the phrase from the message information input by the message input step;
A phrase selection step of selecting a phrase having an attribute related to the media data from the extraction result of the message analysis step;
A metadata generation method comprising: a metadata registration step of storing the selected word and the attribute of the word in the storage unit as the metadata. In a metadata generation program that causes a computer to generate metadata of media data,
In the computer,
Allowing message information related to the media data to be entered,
Extracting the phrase and the attribute of the phrase from the input message information;
Selecting a phrase having an attribute related to the media data from the extraction result;
A metadata generation program characterized in that the selected word and attribute of the word are stored in the storage means as the metadata.

Images