JP4452122B2 - Metadata generation apparatus and metadata generation program - Google Patents

Description

  The present invention relates to a metadata generation apparatus and a metadata generation program that generate metadata describing information of content based on various information included in the content.

In recent years, research for generating metadata describing content (video content) information such as a program from various information included in the content has been performed, and various metadata generation methods have been disclosed. For example, video content is analyzed and its features are used as metadata (see Patent Document 1), keywords are extracted from text information in the content as metadata (see Patent Document 2), and sound is analyzed. In addition, there is a method in which information for each music section or speech section is used as metadata (see Non-Patent Document 1).
In recent years, research on integrated processing of various information such as video and audio, called multi-modal, has been conducted, and a plurality of media, for example, metadata generated by video processing and audio processing. There is a technique for generating new metadata by integrating metadata (see Non-Patent Document 2).
Japanese Patent Laying-Open No. 2004-7535 (FIG. 1) Japanese Patent Laying-Open No. 2003-228569 (FIG. 1) Kenichi Minami and three others "Video indexing using sound information and its applications", IEICE Transactions Vol. J81-D2 No. 3 pp. 529-537 Shigekazu Sasamori, two others "Event extraction from video media by intermodal collaboration", IEICE PRMU 98-189, pp. 99-106 (1999)

Since the above-described conventional metadata generation method generates metadata by a unique method specific to each, the data format of the generated metadata varies depending on the method. Therefore, for example, when generating metadata related to video, audio, etc. from one content, a plurality of metadata having different data formats are generated. Therefore, in order to generate one standardized metadata as content, a plurality of metadata are generated. There is a problem that it takes time and effort to individually edit the metadata.
In addition, tools (devices and programs) for general-purpose generation and editing of individual metadata have not been reported so far, and various types of metadata can be generated uniformly in the same operating environment using a common interface.・ The demand for tools that can be edited is increasing.

  The present invention has been made to solve the above-described problems, and provides a common operating environment for users (users) when generating content metadata, such as video, audio, and the like. It is an object of the present invention to provide a metadata generation apparatus and a metadata generation program that can generate and edit different metadata in a unified format.

  The present invention was devised to achieve the above object, and first, the metadata generation device according to claim 1 represents the content of the content for each event section of a certain event in the content. A metadata generation device for generating section metadata, which is information for each event section in the content, associated with information, section information extraction means, event selection means, section information acquisition processing means, A semantic information extraction unit, a type selection unit, and a semantic information acquisition processing unit are provided.

According to such a configuration, the metadata generation apparatus includes a plurality of section information extraction units that detect an event from the content and extract at least the start time and the end time of the event as section information. By instructing one of the events as a selected event, the section information extraction unit corresponding to the selected event is selected. This enables the user (user) of the metadata generation apparatus to select section information extraction means for extracting section information indicating an event section in which a desired event has occurred from the content. . Here, the event is a unique phenomenon or change that occurs in the content, and is, for example, switching of a video scene, a change in audio frequency, or the like. Further, the event section refers to a section showing features of video, audio, etc., which are contents, such as a section between cut points at which video scenes are switched and an audio section in a certain frequency region.
Then, the metadata generation device acquires the section information of the selected event in the content from the section information extraction unit selected by the event selection unit by the section information acquisition processing unit, and generates first section metadata. As a result, section metadata including at least the start time and end time of the event is generated.

Further, the metadata generation apparatus includes a plurality of semantic information extraction means for analyzing the content and extracting semantic information indicating the contents of the content in association with the start time and the end time in the section information. By instructing one of the types of semantic information as a selection type, a semantic information extraction unit corresponding to the selection type is selected. As a result, the user of the metadata generation apparatus can select semantic information extraction means for extracting desired semantic information from the content. Here, the semantic information is information indicating the meaning and content in a certain section in the content (here, the event section indicated by the section information). For example, the name of the person appearing in the video and its operation, Or the comment etc. which an announcer speaks by a sports broadcast etc.
Then, the metadata generation device acquires the semantic information from the semantic information extraction unit selected by the type selection unit by the semantic information acquisition processing unit, and associates the segment information with the semantic information. Is generated. Thereby, section metadata in which semantic information is associated with each event section is generated.

  The metadata generation apparatus according to claim 2 is the metadata generation apparatus according to claim 1, wherein the section information extraction unit holds a character string indicating a function of the section information extraction unit in advance. The event selection unit obtains the character string from the section information extraction unit, thereby displaying the character string in a list on the screen of the display device.

  According to this configuration, in the metadata generation apparatus, the event selection unit acquires the character string from the section information extraction unit by holding the character string indicating the function of each section information extraction unit in advance. . Then, the metadata generation device displays a list of the character strings on the screen of the display device. Thus, the user can recognize which section information extraction means can be used and which section information extraction means has what function.

  Furthermore, the metadata generation device according to claim 3 is the metadata generation device according to claim 1 or 2, wherein the section information extraction unit extracts the section information in advance by the section information extraction unit. A parameter entry screen, and the section information acquisition processing means acquires the parameter item from the section information extraction means, thereby displaying the parameter input screen displaying the parameter item. It was set as the structure displayed on.

  According to such a configuration, the metadata generation apparatus holds the parameter items for the individual section information extraction unit to extract the section information in advance, so that the section information acquisition processing unit is connected to the section information extraction unit. Get the parameter item. Then, the metadata generation apparatus prompts the user to input parameters by displaying a parameter input screen displaying the parameter items on the display device. Accordingly, the metadata generation apparatus can provide a common interface to the user even when different section information extraction means are operated.

  The metadata generation apparatus according to claim 4 is the metadata generation apparatus according to claim 1, wherein the semantic information extraction unit holds a character string indicating a function of the semantic information extraction unit in advance. The type selection unit obtains the character string from the semantic information extraction unit and displays the character string in a list on the screen of the display device.

  According to this configuration, in the metadata generation device, each type of semantic information extraction unit holds a character string indicating its function in advance, so that the type selection unit acquires the character string from the semantic information extraction unit. . Then, the metadata generation device displays a list of the character strings on the screen of the display device. Thus, the user can recognize which semantic information extracting means can be used and which semantic information extracting means has what function.

  Furthermore, the metadata generation device according to claim 5 is the metadata generation device according to claim 1 or 4, wherein the semantic information extraction unit extracts the semantic information in advance by the semantic information extraction unit. A parameter entry screen, and the semantic information acquisition processing unit acquires the parameter item from the semantic information extraction unit, thereby displaying the parameter input screen displaying the parameter item. It was set as the structure displayed on.

  According to such a configuration, the metadata generation apparatus holds the parameter items for each semantic information extraction unit to extract semantic information in advance, so that the semantic information acquisition processing unit can Get the parameter item. Then, the metadata generation apparatus prompts the user to input parameters by displaying a parameter input screen displaying the parameter items on the display device. Accordingly, the metadata generation apparatus can provide a common interface to the user even when different semantic information extraction means are operated.

  The metadata generation device according to claim 6 is the metadata generation device according to any one of claims 1 to 5, wherein the semantic information acquisition processing unit acquires the semantic information. In addition, section metadata in which a plurality of pieces of semantic information are associated with the section information is generated.

  According to this configuration, the metadata generation apparatus selects one section that is the same event section each time the semantic information acquisition processing means selects different semantic information extraction means by the type selection means. Section metadata in which different semantic information is sequentially added to the information is generated.

  Furthermore, the metadata generation program according to claim 7 extracts a plurality of different events for each event, wherein at least a start time and an end time of the event are extracted from the content as section information for a certain event in the content. Based on the section information extraction module and a plurality of semantic information extraction modules that differ for each type of the semantic information, which analyzes the content and extracts semantic information indicating the content of the content in association with time In order to generate section metadata that is information for each event section, the computer is configured to function as an event selection unit, a section information acquisition processing unit, a type selection unit, and a semantic information acquisition processing unit.

According to such a configuration, the metadata generation program selects an interval information extraction module corresponding to the selected event by instructing one of the plurality of events as a selected event by the event selecting unit. This makes it possible to select a section information extraction module that extracts section information indicating an event section in which an event desired by the user has occurred from the content.
Then, the metadata generation program acquires the section information of the selected event in the content from the section information extraction module selected by the event selection unit by the section information acquisition processing unit, and generates first section metadata.

Furthermore, the metadata generation program selects one of the semantic information extraction modules corresponding to the selected type by instructing one of the types of semantic information as the selected type by the type selection unit. This makes it possible to select a semantic information extraction module that extracts semantic information desired by the user from the content.
The metadata generation program acquires the semantic information from the semantic information extraction module selected by the type selection unit by the semantic information acquisition processing unit, and associates the segment information with the semantic information. Is generated.

  According to the first aspect of the present invention, the metadata generation apparatus can generate section metadata in which section information and semantic information are associated with each other for each event section of the content. At this time, the user can generate the section metadata with a simple operation by simply selecting a means for performing a desired function from the plurality of section information extracting means and semantic information extracting means. Furthermore, the section metadata generated in the metadata generation apparatus has a common format including section information and semantic information even when the functions of the section information extraction unit and the semantic information extraction unit are different. For this reason, the metadata generation device has conventionally become a platform that can handle individually generated metadata in common, processing metadata (section metadata) in a distributed manner, and processing metadata mutually. Various distribution forms such as exchange can be realized.

  According to the invention described in claim 2 or claim 4, the metadata generation device displays a list of character strings indicating functions held by the section information extraction unit or the semantic information extraction unit on the screen. Can select a means for performing a desired function with a simple operation of selecting one of the displayed items. In addition, when adding section information extraction means or semantic information extraction means, section information extraction means or semantic information extraction means can be added simply by adding a function of notifying a character string as a common interface. A character string indicating the function of the extracting means or the semantic information extracting means can be added to the list display.

  According to the invention described in claim 3 or claim 5, the metadata generation device displays the parameter items held by the section information extraction means or the semantic information extraction means on the input screen, and Even when means having different functions are selected in order to prompt the input of parameters, the user can easily recognize and input parameters for realizing the functions. Further, when adding section information extraction means or semantic information extraction means, simply adding a function for notifying parameter items as a common interface, the parameters of the section information extraction means or semantic information extraction means can be easily set. Items can be displayed on the input screen.

  According to the sixth aspect of the present invention, the metadata generation apparatus can associate a plurality of semantic information with one piece of section information, and even in the case of using different semantic information extraction means, sections in a common format Metadata can be generated. Thus, the metadata generation apparatus can function as an integrated platform for generating metadata.

  According to the invention described in claim 7, the metadata generation program can generate section metadata in which section information and semantic information are associated with each other for each event section of the content. At this time, since the metadata generation program uses a modularized program as a means for extracting each section information and semantic information, it is possible to easily add a function by generating a module having a common interface. Can do.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[Configuration of metadata generation system]
First, the configuration of the metadata generation system according to the present invention will be described with reference to FIG. FIG. 1 is a block diagram showing a configuration of a metadata generation system according to the present invention.
As shown in FIG. 1, the metadata generation system 100 extracts information for each section (segment) having a certain time width from content (video, audio, text information) based on its characteristics, By associating meanings (labels) with the sections, metadata (section metadata) for each extracted section is generated and edited. Here, the metadata generation system 100 includes a metadata generation device 1, a module storage device 2, and a content storage device 3.

  The metadata generation apparatus 1 selects a module stored in the module storage apparatus 2, and generates and edits section metadata from the content by incorporating and operating the module in the metadata generation apparatus 1. Is. A detailed configuration of the metadata generation apparatus 1 will be described later.

  The module storage device 2 is a program that extracts information for each section (segment) from the content (module: section information extraction module SM (SM1, SM2, SM3,...)), And a program that associates the meaning (label) with the section. (Module: semantic information extraction module LM (LM1, LM2, LM3...)), And is a storage device such as a hard disk.

  The section information extraction module SM extracts a section serving as the feature as a start time and an end time based on the feature of the content. The section information extraction module SM is selected by a section information extraction module selection unit 10 of the metadata generation apparatus 1 to be described later, and operates by being incorporated as a section information extraction unit 10a of the metadata generation apparatus 1. This section information extraction module SM can be generated as a dynamic link library (DLL), for example. Here, it is assumed that the section information extraction modules (SM1, SM2, SM3,...) Extract section information based on different features.

  For example, when the program (content) is a soccer relay program, the section information extraction module SM1 selects a time section having a certain volume from an audio file (audio A1 in the content storage device 3) digitized from the audience cheers. Acts as an extractor. The section information extraction module SM is not limited to this as long as it extracts a specific target section from the content. For example, the section information extraction module SM detects a cut point that is a video switching frame, and cuts the cut point. A section between points (start time, end time) may be extracted as section information.

  The semantic information extraction module LM extracts the contents of the content as semantic information from the content in association with the segment information extracted by the segment information extraction module SM. The semantic information extraction module LM is selected by the semantic information extraction module selection means 12 of the metadata generation apparatus 1 described later, and is incorporated and operates as the semantic information extraction means 12a of the metadata generation apparatus 1. This semantic information extraction module LM is generated as a dynamic link library (DLL) in the same manner as the section information extraction module SM, and a plurality of semantic information extraction modules (LM1, LM2, LM3...) That extract different semantic information from the same content. ) Is stored in the module storage device 2.

  For example, when the program (content) is a soccer relay program, the semantic information extraction module LM1 uses the section information extraction module SM1 from the video file (video V1 in the content storage device 3) using a general face recognition technique. A player reflected in the video in the extracted section is detected, and the player name is associated with the section information as semantic information. The semantic information extraction module LM is not limited to this as long as the target content is extracted from the content as semantic information. For example, a text in which an announcer's comment in a soccer relay program is converted into text A comment corresponding to the section information may be extracted from the information (text information T1 in the content storage device 3).

The content storage device 3 stores content and is a storage device such as a hard disk. Here, the content storage device 3 stores video (V1, V2,...), Audio (A1, A2,...), And text information (T1, T2,...) As content.
Here, the video (V1 or the like) is, for example, a video file such as MPEG. The audio (A1 etc.) is an audio file containing audio at the same time as the video (V1 etc.). For example, the audio A1 is an audio file in a standard WAVE format (audio data recorded by PCM) of Windows (registered trademark). The text information (T1 etc.) is a text file in which various information associated with the playback time of the video (V1 etc.) is described as text data. For example, the text information T1 is a text file in which an announcer's comment is converted into text for each time.

  As described above, the metadata generation apparatus 1 selects a module (section information extraction module SM) for extracting section information from the module storage apparatus 2, and divides (segments) content into sections by the module. And the metadata production | generation apparatus 1 selects the module (semantic information extraction module LM) which extracts semantic information from the module storage apparatus 2, and associates semantic information with the section information segmented by the module, Generate section metadata.

  Here, the section metadata will be described with reference to FIG. 2A and 2B are explanatory diagrams for explaining the configuration of the section metadata. FIG. 2A shows the structure of the section metadata, and FIG. 2B shows an example of the section metadata.

As shown in FIG. 2A, the section metadata is data indicating the content content including identification information, section information, and semantic information.
Here, the identification information is unique information for identifying the section metadata, and is an identifier (ID) such as a numerical value.
The section information is information indicating a time section of the section metadata in the content, and is a start time (In) and an end time (Out).

  The semantic information is information in which the meaning / content of the content segmented by the segment information is associated with a key (Key) and a value (Value). Here, the key (Key) is information indicating the type of meaning / content. The value (Value) is detailed information of the type indicated by the key (Key). Furthermore, here, information indicating the relationship (interval relationship) of each section metadata is added as semantic information. In this semantic information, the section relationship is associated with a type (Type) and a target (Target). Here, the type (Type) is information indicating what meaning the other sections have with respect to the section. The target is information (identifier) indicating another section that is a target of the section relationship. It is possible to associate a plurality of semantic information in the same section metadata.

A specific example of the section metadata will be described with reference to FIG.
As shown in FIG. 2B, the section metadata M1, the section metadata M2, and the section metadata M3 are identified by the identifier ID (“001”, “002”, “003”) that is identification information.
In addition, the section metadata indicates which time section is the metadata at the content time (program time) by the start time In and the end time Out. For example, the section metadata M1 represents metadata of a time section from “09: 0: 32-00” to “09: 0: 54-00”. The time “XX: XX: XX-XX” indicates “hour: minute: second-frame number”.

  The section metadata indicates the contents of the contents in the section by using a key (Key) and a value (Value) indicating the meaning / content. For example, the section metadata M1 uses the “event” indicating that an event has occurred as a key (Key), and “shoot”, which is detailed information of the event, as a value (Value). Is shown. Further, the section metadata M2 uses “player” indicating that a player is reflected in the video as a key (Key), and the player name “B player”, which is detailed information of the player, as a value (Value). Indicates the meaning and content of the section.

Furthermore, the section metadata indicates a relationship between the section and another section by a type (Type) indicating the section relationship and a target (Target). For example, the section metadata M1 indicates that another section (here, the identifier “003”) is a target (Target), and the video of the other section is the slow playback video (“slow playback”) of this section. , By type.
In this way, by defining the section metadata, even if section information is extracted from the content by different features or different semantic information is extracted, it is expressed in the form of a common section metadata. can do.
Hereinafter, referring back to FIG. 1, the detailed configuration of the metadata generation apparatus will be described.

[Configuration of metadata generation device]
As shown in FIG. 1, the metadata generation device 1 includes a section information extraction module selection unit 10, a section information acquisition processing unit 11, a semantic information extraction module selection unit 12, a semantic information acquisition processing unit 13, and a section metadata. A data generation control unit 14, a metadata storage unit 15, a selected video reproduction unit 16, a section metadata editing unit 17, a user instruction receiving unit 18, and a section metadata input / output unit 19 are provided.

  The section information extraction module selection means (event selection means) 10 selects one of a plurality of events as a selection event (for example, cheer of the audience in a soccer broadcast program) via a user instruction reception means 18 described later. The section information extraction module SM corresponding to the selected event is selected from the module storage device 2. The section information extraction module selection unit 10 acquires character strings indicating individual functions from the plurality of section information extraction modules SM, and presents them to the user via the user instruction receiving unit 18 to Information for assisting selection of the section information extraction module SM is presented.

  Then, the section information extraction module selection unit 10 expands and operates the selected section information extraction module SM in a memory (not shown). Thereby, the section information extraction module SM is caused to function as the section information extraction means 10a. The section information extraction means (section information extraction module) 10a operates based on an instruction from the section information acquisition processing means 11 described later.

  Here, the section information extraction means 10a shows one module selected in the plurality of section information extraction modules SM, but the metadata is obtained by using the plurality of section information extraction modules SM as a plurality of section information extraction means 10a in advance. It is good also as incorporating in the production | generation apparatus 1 and comprising. However, when the section information extraction module is configured to be provided as a general module, a mode in which the user operates by selecting a necessary module and incorporating it in the metadata generation apparatus 1 as in this embodiment. This is preferable because of its high versatility.

  The section information acquisition processing unit 11 operates the section information extraction unit (section information extraction module) 10a selected by the section information extraction module selection unit 10 to operate the section information of the selected event in the content from the section information extraction unit 10a. To generate section metadata (first section metadata). Here, the section information acquisition processing unit 11 includes a parameter setting unit 11a and a section information acquisition unit 11b.

The parameter setting unit 11a obtains parameter items necessary for extracting section information from content from the section information extracting unit 10a, and sets parameter values input by the user in the section information extracting unit 10a. is there. The parameter setting unit 11a presents the parameter items acquired from the section information extraction unit 10a to the user via the user instruction receiving unit 18. Then, the parameter setting unit 11a acquires the parameter value input by the user from the user instruction receiving unit 18 and sets it in the section information extracting unit 10a. Here, the parameter necessary for extracting the section information is, for example, a file name of an audio file indicating from which voice (A1 or the like) the section information is extracted.
In this way, by acquiring necessary parameter items from the section information extraction unit 10a and presenting them to the user, even if the section information extraction unit 10a is a different section information extraction module SM, an operation interface for the user is provided. Can be unified.

  The section information acquisition unit 11b instructs the section information extraction unit 10a to extract section information, causes the section information extraction unit 10a to perform the section information extraction operation, and acquires the section information as a result of the extraction. It is. The section information acquired here is written in the metadata storage unit 15 as section metadata (M1, M2,...). By the section information acquisition unit 11b, only the identifier ID, which is identification information shown in FIG. 2, and the start time In and end time Out, which are section information, are written in the section metadata (M1, M2,...). Note that the first section metadata described in the claims is metadata in which only section information is described in association with identification information.

  The semantic information extraction module selection unit (type selection unit) 12 selects one of a plurality of analysis method types for analyzing semantic information via a user instruction reception unit 18 described later (for example, face recognition analysis, announcer). As a comment analysis), the semantic information extraction module LM corresponding to the selected type is selected from the module storage device 2 when instructed by the user. The semantic information extraction module selection unit 12 acquires character strings indicating individual functions from a plurality of semantic information extraction modules LM, and presents them to the user via the user instruction reception unit 18, thereby providing the user with a character string. Information for assisting selection of the semantic information extraction module LM is presented.

  Then, the semantic information extraction module selection unit 12 develops and operates the selected semantic information extraction module LM in a memory or the like (not shown). As a result, the semantic information extraction module LM functions as the semantic information extraction means 12a. The semantic information extraction means (meaning information extraction module) 12a operates based on an instruction from the semantic information acquisition processing means 13 described later.

  Here, the semantic information extraction unit 12a indicates one module selected in the plurality of semantic information extraction modules LM, but the metadata is preliminarily defined by using the plurality of semantic information extraction modules LM as the plurality of semantic information extraction units 12a. It is good also as incorporating in the production | generation apparatus 1 and comprising. However, when the semantic information extraction module is provided in a form provided as a general module, a form in which the user operates by selecting a necessary module and incorporating it in the metadata generation apparatus 1 as in this form. This is preferable because of its high versatility.

  The semantic information acquisition processing unit 13 operates the semantic information extraction unit (semantic information extraction module) 12a selected by the semantic information extraction module selection unit 12 to extract from the semantic information extraction unit 12a by the selected analysis method. The acquired semantic information is acquired, and section metadata (second section metadata) is generated. Here, the semantic information acquisition processing unit 13 includes a parameter setting unit 13a and a semantic information acquisition unit 13b.

  The parameter setting unit 13a acquires parameter items necessary for extracting semantic information from the content from the semantic information extracting unit 12a, and sets the parameter value input by the user in the semantic information extracting unit 12a. is there. The parameter setting unit 13a presents the parameter items acquired from the semantic information extracting unit 12a to the user via the user instruction receiving unit 18. Then, the parameter setting unit 13a acquires the parameter value input by the user from the user instruction receiving unit 18 and sets it in the semantic information extracting unit 12a. Here, the parameters necessary for extracting semantic information include, for example, the video file name indicating which video (V1 etc.) the semantic information is to be extracted from, the facial features when performing face recognition, and the like. For example, the file name of the text file associated with the player name.

In this way, by acquiring necessary parameter items from the semantic information extracting unit 12a and presenting them to the user, even if the semantic information extracting unit 12a is a different semantic information extracting module LM, The input screen can be displayed, and the operation interface for the user can be unified.
The section information (start time, end time) of the section metadata (M1, M2,...) Stored in the metadata storage unit 15 is set in the semantic information extraction unit 12a as a parameter as it is.

  The semantic information acquisition unit 13b instructs the semantic information extraction unit 12a to extract semantic information, causes the semantic information extraction unit 12a to perform the semantic information extraction operation, and acquires the semantic information as a result of the extraction. It is. The semantic information acquired here is associated with the section information (start time, end time) of the section metadata (M1, M2,...) Stored in the metadata storage unit 15 in association with the section metadata (M1, M2,...). ) To be written. As a result, identification information, section information, and semantic information are written in the section metadata (M1, M2,...). Note that the second section metadata described in the claims is metadata in which section information and semantic information are described in association with identification information.

  The section metadata generation control unit 14 is a main control unit that controls the entire metadata generation apparatus 1. The section metadata generation control unit 14 determines an operation based on the control information notified from each unit, and gives instructions to other units. For example, when the section metadata generation control unit 14 receives an instruction from the section information extraction module 10 to present a character string indicating the function of the section information extraction module SM to the user, the section metadata generation control unit 14 Thus, an instruction to present a selection screen for selecting the section information extraction module SM is given to a display device (not shown).

The metadata accumulating unit 15 accumulates section metadata and is a storage device such as a semiconductor memory or a hard disk. The section metadata (M1, M2,...) Stored in the metadata storage unit 15 is generated by writing section information and semantic information by the section information acquisition processing unit 11 and the semantic information acquisition processing unit 13. .
The format of the section metadata stored here is not particularly limited, but is, for example, an XML (Extensible Markup Language) format such as MPEG-7.

  The selected video playback means 16 reads the content video corresponding to the section metadata selected by the user from the content storage device 3 via the user instruction receiving means 18 in the section metadata (M1, M2,...). To play. For example, when the video (V1 or the like) stored in the content storage device 3 is an MPEG-encoded video, the selected video playback means 16 functions as an MPEG decoder and plays back the video. Display video on no display device. As a result, the user can confirm whether or not the selected section metadata is normally segmented.

  The section metadata editing unit 17 presents the section metadata (M1, M2,...) To the user via the user instruction receiving unit 18, and corrects the section metadata. For example, the section metadata editing unit 17 presents the contents of the section metadata (section information and semantic information) on the display screen via the user instruction receiving unit 18, and the user uses a keyboard or the like (not shown) to When the content of the section metadata is corrected, the content of the metadata storage unit 15 is updated based on the correction result.

  The user instruction accepting means 18 presents a screen for accepting an instruction from the user on a display device (not shown), and accepts an instruction from the user via an input device such as a mouse or a keyboard (not shown). . The content instructed here is notified to the section metadata generation control means 14 as needed, and an operation according to the content of the instruction is executed. Note that a screen for receiving an instruction from the user will be described later by showing a screen example.

The section metadata input / output means 19 is an interface for inputting / outputting section metadata. For example, it is a disk drive device or the like that writes section metadata to or reads section metadata from a recording medium such as a DVD-RAM. Alternatively, it may be a communication board that transmits / receives section metadata via a communication line.
Here, since the section metadata (M1, M2,...) Has a common data format, the section metadata can be used by inputting the section metadata generated externally. . For example, it is possible to input section metadata in which only section information is written by another metadata generation apparatus, and the metadata generation apparatus 1 can add semantic information to the section metadata.

  The configuration of the metadata generation apparatus 1 has been described above, but the present invention is not limited to this. For example, the metadata generation device 1 can be realized by causing a general computer to execute a program and operating an arithmetic device or a storage device in the computer. This program (metadata generation program) can be distributed via a communication line, or can be distributed by writing in a recording medium such as a CD-ROM.

[Example screen of metadata generation device]
Next, an example of a screen displayed by the metadata generation apparatus 1 will be described.
(Section information extraction screen)
First, referring to FIGS. 3 and 4 (refer to FIG. 1 as appropriate), an example of a screen for selecting the section information extraction module SM and extracting section information from the content will be described. FIG. 3 is a screen example showing an example of a selection screen for selecting a section information extraction module. FIG. 4 is a screen example showing an example of a parameter input screen for inputting parameters to be notified to the section information extraction module.

As shown in FIG. 3, the selection screen W1 for selecting the section information extraction module SM displays a list of character strings indicating the functions of the section information extraction module SM (list display), and receives a selection from the user. It has become.
In the metadata generation apparatus 1, when the section information extraction module selection unit 10 selects the section information extraction module SM, the selection screen W <b> 1 is displayed on the screen via the user instruction reception unit 18. When the list button B is clicked with the mouse on the selection screen W1, the section information extraction module selection means 10 acquires the individual section information extraction modules (SM1, SM2, SM3) acquired from the plurality of section information extraction modules SM. Display a list of character strings indicating the function of (...). Here, the character strings “cheer segmentation module (SL1)” and “video cut point segmentation module (SL2)” are acquired from the section information extraction module SM1, and the character strings SL1 and SL2 are acquired from the section information extraction module SM2. Is displayed as a list. Accordingly, the user can select what function is used to extract the section information.
Then, when the section information extraction module SM is selected on the selection screen W1, the parameter input screen W2 of FIG. 4 is displayed.

As shown in FIG. 4, the input screen W2 for inputting the parameter values necessary for extracting the section information is provided with input fields AR1 and AR2 for the parameters P1 and P2, respectively. The screen accepts input.
In the metadata generation device 1, the parameter setting unit 11 a of the section information acquisition processing unit 11 acquires parameters necessary for extracting section information from the section information extraction unit (section information extraction module) 10 a, The input screen W2 is displayed on the screen via the accepting means 18. This input screen W2 shows a state in which “analysis target file name (P1)” and “maximum entropy method model file name (P2)” are notified from the section information extraction unit 10a as parameters. Thereby, the user inputs the value (file name) of each parameter in the input fields AR1 and AR2 of each parameter and clicks the execution button EX with the mouse, thereby extracting the section information.

(Semantic information extraction screen)
Next, referring to FIGS. 5 and 6 (refer to FIG. 1 as appropriate), an example of a screen for selecting the semantic information extraction module LM and extracting semantic information from the content will be described. FIG. 5 is a screen example showing an example of a selection screen for selecting the semantic information extraction module. FIG. 6 is a screen example showing an example of a parameter input screen for inputting a parameter to be notified to the semantic information extraction module.

As shown in FIG. 5, the selection screen W3 for selecting the semantic information extraction module LM displays a list of character strings indicating the functions of the semantic information extraction module LM (list display), and receives a selection from the user. It has become.
In the metadata generation apparatus 1, when the semantic information extraction module selection unit 12 selects the semantic information extraction module LM, the selection screen W3 is displayed on the screen via the user instruction reception unit 18. On this selection screen W3, when the list button B is clicked with the mouse, the semantic information extraction module selection means 12 acquires the individual semantic information extraction modules (LM1, LM2, LM3) acquired from the plurality of semantic information extraction modules LM. Display a list of character strings indicating the function of (...). Here, each character string of “face recognition labeling module (LL1)” from the semantic information extraction module LM1 and “comment analysis labeling module (LL2)” from the semantic information extraction module LM2 is acquired, and the character strings LL1, LL2 are acquired. Is displayed as a list. Thereby, the user can select what function the semantic information is extracted from.
Then, when the semantic information extraction module LM is selected on the selection screen W3, the parameter input screen W4 in FIG. 6 is displayed.

As shown in FIG. 6, the input screen W4 for inputting parameter values necessary for extracting semantic information is provided with input fields AR3 and AR4 for the parameters P3 and P4. The screen accepts input.
In the metadata generation apparatus 1, the parameter setting unit 13 a of the semantic information acquisition processing unit 13 acquires parameters necessary for extracting semantic information from the semantic information extraction unit (semantic information extraction module) 12 a, An input screen W4 is displayed on the screen via the accepting means 18. This input screen W4 shows a state in which “analysis target file name (P3)” and “player name file name (P4)” are notified from the semantic information extraction means 12a as parameters. Thereby, the user inputs the value (file name) of each parameter in the input field AR3, AR4 of each parameter, and the semantic information is extracted by clicking the execution button EX with the mouse.

(Section metadata confirmation / edit screen)
Next, referring to FIG. 7 (refer to FIG. 1 as appropriate), a screen example for confirming / editing the generated section metadata on the screen will be described. FIG. 7 is a screen example showing an example of a confirmation / editing screen for confirming / editing the section metadata.

As shown in FIG. 7, the confirmation / editing screen W5 includes a video reproduction area Va, a section information presentation area SI, a semantic information (meaning / content) presentation area LI1, and a semantic information (section relation) presentation area LI2. The screen is mainly composed. Here, an information presentation area Sa for presenting information from the section information extraction module SM and an information presentation area La for presenting information from the semantic information extraction module LM are further configured on the screen.
Since the section metadata is uniquely determined by the section information, the section metadata is selected here by selecting the section information displayed in the section information presentation area SI.

  The video playback area Va is an area for playing back and displaying a video of a time section indicated by section information (start time In and end time Out) of the selected section metadata. In the metadata generation device 1, the user instruction receiving means 18 detects that the section metadata has been selected and notifies the selected video playback means 16 so that the selected video playback means 16 can display the video in the corresponding time section. Play.

  The section information presentation area SI is an area for displaying identification information (identifier ID) extracted from the section metadata and section information (start time In, end time Out). Here, the frame image (Fa1 and the like) at the start time In is displayed simultaneously. The section metadata is selected by the user selecting desired section information Se in the section information presentation area SI with the mouse.

  The semantic information (meaning / content) presentation area LI1 is an area for displaying keys and values that are semantic information (meaning / content) extracted from the section metadata. In the semantic information (meaning / content) presentation area LI1, semantic information (meaning / content) of the section metadata is displayed when the section metadata is selected in the section information presentation area SI.

The semantic information (section relation) presentation area LI2 is an area for displaying a type and target that are semantic information (section relation) extracted from the section metadata. In the semantic information (section relation) presentation area LI2, semantic information (section relation) of the section metadata is displayed when the section metadata is selected in the section information presentation area SI.
The section information presentation area SI, the semantic information (meaning / content) presentation area LI1, and the semantic information (section relation) presentation area LI2 are displayed by the section metadata editing means 17, and can be edited from a keyboard or the like. By inputting data, section information and semantic information are changed.

  The information presentation area Sa is an area for presenting detailed information when the detailed information can be acquired from the section information extraction module SM. This information presentation area Sa presents a character string when the extraction method, version, etc. of the section information extraction module SM can be acquired as a character string from the section information extraction module SM. Alternatively, information other than the information presented in the section information presentation area SI, for example, color average, maximum volume, minimum volume, etc. may be presented as the section information.

The information presentation area La is an area for presenting detailed information when the detailed information can be acquired from the semantic information extraction module LM. This information presentation area La presents a character string when the extraction method, version, etc. of the semantic information extraction module LM can be acquired as a character string from the semantic information extraction module LM. Alternatively, as the semantic information, information other than the information presented in the semantic information (meaning / content) presentation area LI1 and the semantic information (section relation) presentation area LI2, for example, the full text of the announcer comment in the section may be presented. Good.
Thus, by visualizing the section metadata, the contents of the section metadata can be easily confirmed and edited.

[Operation of metadata generation device]
Next, the operation of the metadata generation apparatus according to the present invention will be described with reference to FIGS. 8 to 10 (refer to FIG. 1 as appropriate for the configuration). FIG. 8 shows the overall operation of the metadata generation apparatus. It is a flowchart. FIG. 9 is a flowchart showing the section information extraction operation of the metadata generation apparatus. FIG. 10 is a flowchart showing the semantic information extraction operation of the metadata generation apparatus.

(Overall operation)
First, the overall operation of the metadata generation apparatus 1 will be described with reference to FIG. In addition, in FIG. 8, the content of the section metadata produced | generated in each operation | movement is also illustrated.
First, the metadata generation device 1 starts from the content (video (V1 etc.), audio (A1 etc.), text information (T1 etc.)) stored in the content storage device 3, and is the start time and end which are section information Time extraction is performed (step S10).
In this step S10, for example, by extracting section information (segmentation by cheer) from voice (A1 etc.), section metadata including start time and end time as shown in FIG. M1 ₁ , M2 ₁ , M3 ₁ ) are generated. In the drawing, identification information of each section metadata is omitted. The detailed operation of step S10 will be described later with reference to FIG.

Subsequently, the metadata generation device 1 uses the section metadata (M1 ₁ , M2) from the content (video (V1 etc.), audio (A1 etc.), text information (T1 etc.)) stored in the content storage device 3. ₁ , semantic information corresponding to the start time and end time in M3 ₁ ) is extracted (step S20).
In this step S20, for example, by extracting semantic information (labeling by face recognition) from the video (V1 etc.), as shown in FIG. 8B, the section metadata (M1) in FIG. ₁ , M2 ₁ , M3 ₁ ), and section metadata (M1 ₂ , M2 ₂ , M3 ₂ ) in which a key (player) and a value (A player, etc.) are added are generated.
In step S20, semantic information is extracted from the text information (T1 etc.) (labeled by the announcer's comment), so that as shown in FIG. Section metadata (M1 ₃ , M2 ₃ , M3 ₃ ) in which a key (event) and a value (shoot, etc.) are added to the metadata (M1 ₂ , M2 ₂ , M3 ₂ ) is generated.

Then, the metadata generation apparatus 1 outputs the section metadata (M1 ₃ , M2 ₃ , M3 ₃ ) generated in step S20 (step S30).
As described above, the metadata generation apparatus 1 generates and outputs the section metadata by the two steps of extracting the section information from the content (step S10) and extracting the semantic information (step S20). .

(Section information extraction operation)
Next, the operation of the section information extraction operation (step S10 in FIG. 8) in the metadata generation apparatus 1 will be described in detail with reference to FIG.
First, the metadata generation apparatus 1 acquires character strings indicating individual functions from the section information extraction module SM by the section information extraction module selection unit 10 (step S100). Here, the section information extraction module SM returns a character string by a function (IsSegmentModule ()).

  The section information extraction module selection unit 10 that has acquired the character string presents a selection screen (see FIG. 3) for selecting the section information extraction module SM via the user instruction reception unit 18 (step S101). Here, the user instruction accepting means 18 accepts the selection of the section information extraction module SM by the user on the selection screen, and notifies the section information extraction module selection means 10 of the selected section information extraction module SM.

  The metadata generation apparatus 1 incorporates the section information extraction module SM notified from the user instruction receiving unit 18 by the section information extraction module selection unit 10 as the section information extraction unit 10a. As a result, the section information extraction module SM that actually extracts the section information is selected (step S102).

  Then, the metadata generation apparatus 1 uses the parameter setting unit 11a of the section information acquisition processing unit 11 to set parameters (parameter items) necessary for extracting the section information to the section information extracting unit 10a (here, section information extraction). Obtained from the module SM1) (step S103). Here, the section information extraction module SM returns the parameter item by a function (GetRequiredData ()).

  The parameter setting unit 11a that has acquired the parameter presents an input screen (see FIG. 4) for inputting the parameter value via the user instruction receiving unit 18 (step S104). Here, the user instruction accepting means 18 accepts input of the parameter value by the user on the input screen, and notifies the parameter setting means 11a of the inputted parameter value.

  Then, the parameter setting unit 11a notifies (sets) the parameter value to the section information extraction unit 10a (here, the section information extraction module SM1) (step S105). Here, the section information extraction module SM sets parameters using a function (SetRequiredData ()).

  Then, the section information acquiring unit 11b instructs the section information extracting unit 10a to extract section information, thereby executing section information extraction (step S106). Here, the section information extraction module SM returns a section information extraction result by a function (GetSegment ()). The section information extracted here is presented by the section metadata editing means 17 on the confirmation / editing screen (see FIG. 7).

  At this stage, the user confirms the contents of the section information. That is, the metadata generation apparatus 1 uses the user instruction receiving means 18 to re-extract the section information from the user (step S107), edit the section information (step S108), and execute another section information extraction module SM (step S109). ) Is determined whether or not an instruction to perform is input.

When the metadata generation apparatus 1 is instructed to re-extract the section information by the user (Yes in step S107), the metadata generation apparatus 1 returns to step S103 and re-enters the parameter value, thereby re-entering the section information. Perform extraction.
When the metadata generation apparatus 1 is instructed to edit the section metadata by the user (Yes in step S108), the section metadata editing unit 17 edits the section metadata (step S110). After editing the section metadata, the process returns to step S107.

Furthermore, when the metadata generation apparatus 1 is instructed by the user to execute another section information extraction module SM (Yes in step S109), the metadata generation apparatus 1 returns to step S100 and starts the operation from the selection operation of the section information extraction module SM. Try again.
And the metadata production | generation apparatus 1 complete | finishes the extraction operation of area information in the step (No in step S107, step S108, and step S109) in which the user confirmed the content of area information.

(Semantic information extraction operation)
Next, the operation of the semantic information extraction operation (step S20 in FIG. 8) in the metadata generation apparatus 1 will be described in detail with reference to FIG.
First, the metadata generation apparatus 1 acquires character strings indicating individual functions from the semantic information extraction module LM by the semantic information extraction module selection unit 12 (step S200). Here, the semantic information extraction module LM returns a character string by a function (IsLabelModule ()).

  The semantic information extraction module selection unit 12 that has acquired the character string presents a selection screen (see FIG. 5) for selecting the semantic information extraction module LM via the user instruction reception unit 18 (step S201). Here, the user instruction accepting unit 18 accepts selection of the semantic information extraction module LM by the user on the selection screen, and notifies the semantic information extraction module selection unit 12 of the selected semantic information extraction module LM.

  Then, the metadata generation apparatus 1 incorporates the semantic information extraction module LM notified from the user instruction reception unit 18 by the semantic information extraction module selection unit 12 as the semantic information extraction unit 12a. As a result, the semantic information extraction module LM that actually extracts the semantic information is selected (step S202).

Then, the metadata generation apparatus 1 uses the parameter setting unit 13a of the semantic information acquisition processing unit 13 to extract parameters (parameter items) necessary for extracting the semantic information from the semantic information extraction unit 12a (here, semantic information extraction). Obtained from the module LM1) (step S203). Here, the semantic information extraction module LM returns a parameter item by a function (GetRequiredData ()).
The parameter setting unit 13a that has acquired the parameter presents an input screen (see FIG. 6) for inputting the parameter value via the user instruction receiving unit 18 (step S204). Here, the user instruction accepting means 18 accepts the input of the parameter value by the user on the input screen, and notifies the parameter setting means 13a of the inputted parameter value.

  Then, the parameter setting unit 13a notifies (sets) the parameter value to the semantic information extraction unit 12a (here, the semantic information extraction module LM1) (step S205). In this case, the semantic information extraction module LM sets parameters using a function (SetRequiredData ()).

  Then, the semantic information acquisition unit 13b instructs the semantic information extraction unit 12a to extract semantic information, thereby executing the semantic information extraction (step S206). Here, the semantic information extraction module LM returns the extraction result of the semantic information (meaning / content) by the function (GetLabel ()), and returns the extraction result of the semantic information (section relation) by the function (GetRelation ()). I am going to do that. The semantic information extracted here is presented on the confirmation / editing screen (see FIG. 7) by the section metadata editing means 17.

At this stage, the user confirms the content of the semantic information. That is, the metadata generation apparatus 1 uses the user instruction receiving unit 18 to re-extract semantic information from the user (step S207), edit semantic information (step S208), and execute another semantic information extraction module LM (step S209). ) Is determined whether or not an instruction to perform is input.
When the metadata generation apparatus 1 is instructed by the user to redo the extraction of the semantic information (Yes in Step S207), the metadata generation apparatus 1 returns to Step S203 and performs the re-input of the parameter value, thereby performing the semantic information again. Perform extraction.

Further, when the metadata generation apparatus 1 is instructed to edit the semantic information by the user (Yes in step S208), the metadata generation apparatus 1 edits the semantic information by the section metadata editing unit 17 (step S210). After editing this semantic information, the process returns to step S207.
Further, when the metadata generation apparatus 1 is instructed by the user to execute another semantic information extraction module LM (Yes in step S209), the metadata generation apparatus 1 returns to step S200 and starts the operation from the selection operation of the semantic information extraction module LM. Try again.
And the metadata production | generation apparatus 1 complete | finishes the extraction operation | movement of semantic information in the stage (No in step S207, step S208, and step S209) which the user confirmed the content of the semantic information.

  As above, the operation of generating the section metadata from the content stored in advance in the content storage device 3 (offline operation) has been described as the operation of the metadata generation device 1. However, according to the present invention, it is possible to extract section metadata from content input as stream data. Hereinafter, an operation (online operation) for generating section metadata from content input as stream data in the metadata generation apparatus 1 will be described with reference to FIG. 11 (refer to FIG. 1 as appropriate).

(Section metadata generation operation by online processing)
FIG. 11 is a flowchart showing the section metadata generation operation by online processing of the metadata generation apparatus.
First, the metadata generation apparatus 1 is necessary for selecting the section information extraction module SM by the section information extraction module selection unit 10 and extracting the section information by the parameter setting unit 11a of the section information acquisition processing unit 11. A parameter is set (step S40). This step S40 corresponds to step S100 to step S105 in FIG.

Further, the metadata generation apparatus 1 is necessary for selecting the semantic information extraction module LM by the semantic information extraction module selecting unit 12 and extracting the semantic information by the parameter setting unit 13 a of the semantic information acquisition processing unit 13. A parameter is set (step S41). This step S41 corresponds to step S200 to step S205 in FIG.
Then, the metadata generation apparatus 1 extracts one piece of section information (one segment) from the input content through the section information extraction unit 10a by the section information acquisition unit 11b of the section information acquisition processing unit 11 (Step 1). S42).

  Subsequently, the metadata generation apparatus 1 uses the semantic information acquisition unit 13b of the semantic information acquisition processing unit 13 within the section (in one segment) indicated by the section information extracted in step S42 via the semantic information extraction unit 12a. ) To extract semantic information from the content (step S43). The content data of the section indicated by the section information is temporarily stored in storage means (not shown), and the semantic information is extracted from the content data in the section by the semantic information extraction means 12a.

When the section metadata is checked and edited for each section (Yes in step S44), the section metadata editing means 17 extracts the section metadata that is extracted in step S43 and displays the generated section metadata on the display screen. Then, confirmation / editing work is performed (step S45).
Furthermore, when outputting section metadata for every section (Yes in step S46), the section metadata input / output means 19 outputs section metadata for one segment (step S47).

Then, it is determined whether or not the content input has been completed (step S48). If it has not been completed (No in step S48), the process returns to step S42 to continue extracting the section information and the semantic information.
On the other hand, when the input of the content is completed (Yes in step S48), it is determined whether or not to confirm / edit the section metadata of all the sections (step S49), and when confirmation / editing is performed (step S49). Yes), the section metadata editing means 17 displays the confirmation / editing screen as shown in FIG. 7 and performs the confirmation / editing work (step S50).
And the metadata production | generation apparatus 1 outputs the produced | generated area metadata (for several segments) by the area metadata input / output means 19 (step S51).
Through the above operation, the metadata generation apparatus 1 can generate and output the section metadata from the content by online processing.

It is a block diagram which shows the structure of the metadata production | generation system containing the metadata production | generation apparatus which concerns on this invention. It is explanatory drawing for demonstrating the structure of area metadata, Comprising: (a) is a structure of area metadata, (b) shows the example of area metadata. It is an example of a screen which shows an example of the selection screen which selects an area information extraction module. It is an example of a screen which shows an example of the parameter input screen which inputs the parameter notified to an area information extraction module. It is an example of a screen which shows an example of the selection screen which selects a semantic information extraction module. It is a screen example which shows an example of the parameter input screen which inputs the parameter notified to a semantic information extraction module. It is an example of a screen which shows an example of the confirmation / edit screen which confirms and edits section metadata. It is a flowchart which shows the whole operation | movement of the metadata production | generation apparatus which concerns on this invention. It is a flowchart which shows the extraction operation | movement of the section information of the metadata production | generation apparatus which concerns on this invention. It is a flowchart which shows the extraction operation | movement of the semantic information of the metadata production | generation apparatus which concerns on this invention. It is a flowchart which shows the area metadata production | generation operation | movement by the online process of the metadata production | generation apparatus which concerns on this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Metadata production | generation apparatus 10 Section information extraction module selection means (event selection means)
10a Section information extraction means 11 Section information acquisition processing means 11a Parameter setting means 11b Section information acquisition means 12 Semantic information extraction module selection means (type selection means)
12a Semantic information extraction means 13 Semantic information acquisition processing means 13a Parameter setting means 13b Semantic information acquisition means 14 Section metadata generation control means 15 Metadata storage means 16 Selected video reproduction means 17 Section metadata editing means 18 User instruction reception means 19 Section Metadata input / output means SM Section information extraction module LM Semantic information extraction module

Claims (7)

In the content, for each event section of a certain event, a metadata generation device that generates section metadata that is information for each event section in the content in association with semantic information indicating the content of the content,
A plurality of section information extraction means different for each event, detecting the event from the content, and extracting at least a start time and an end time of the event as section information;
An event selection unit that selects the section information extraction unit corresponding to the selected event by instructing one of the plurality of events as a selected event;
Section information acquisition processing means for acquiring section information of the selected event in the content from the section information extracting means selected by the event selecting means, and generating first section metadata;
Analyzing the content and extracting the semantic information in association with a start time and an end time in the section information, and a plurality of different semantic information extraction means for each type of the semantic information;
By selecting one of the types of semantic information as a selection type, a type selection unit that selects the semantic information extraction unit corresponding to the selection type;
Semantic information acquisition processing means for acquiring the semantic information from the semantic information extraction means selected by the type selection means, and generating second section metadata in which the section information and the semantic information are associated with each other;
A metadata generation apparatus comprising:   The section information extraction unit holds a character string indicating the function of the section information extraction unit in advance, and the event selection unit acquires the character string from the section information extraction unit, thereby the character string The metadata generation device according to claim 1, wherein a list is displayed on a screen of the display device.   The section information extraction means holds in advance the parameter items for extracting the section information in the section information extraction means, and the section information acquisition processing means extracts the parameter items from the section information extraction means. The metadata generation apparatus according to claim 1 or 2, wherein an acquisition screen for displaying the parameter item is displayed on the display device by acquiring the parameter item.   The semantic information extraction unit holds a character string indicating the function of the semantic information extraction unit in advance, and the type selection unit acquires the character string from the semantic information extraction unit, thereby The metadata generation apparatus according to claim 1, wherein a list is displayed on a screen of the display apparatus.   The semantic information extraction means holds in advance parameters items for extracting the semantic information in the semantic information extraction means, and the semantic information acquisition processing means extracts the parameter items from the semantic information extraction means. 5. The metadata generation apparatus according to claim 1, wherein an acquisition screen of the parameter displaying the parameter item is displayed on the display device.   6. The semantic data acquisition processing unit generates section metadata in which a plurality of the semantic information is associated with the section information every time the semantic information is acquired. The metadata generation device according to claim 1. In a content, for a certain event, at least a start time and an end time of the event are extracted from the content as segment information, a plurality of segment information extraction modules different for each event, the content is analyzed, and the content Generating section metadata that is information for each event section in the content based on a plurality of semantic information extraction modules that are different for each type of the semantic information, and extracts semantic information indicating the content of the content in association with time For the computer,
Event selection means for selecting the section information extraction module corresponding to the selected event by instructing one of the plurality of events as a selected event,
Section information acquisition processing means for acquiring section information of the selected event in the content from the section information extraction module selected by the event selecting means, and generating first section metadata.
Type selection means for selecting the semantic information extraction module corresponding to the selected type by instructing one of the types of semantic information as a selected type,
Semantic information acquisition processing means for acquiring the semantic information from the semantic information extraction module selected by the type selection means, and generating second section metadata in which the section information and the semantic information are associated with each other;
A metadata generation program characterized by functioning as

Images