JP2006343941A - Content retrieval/reproduction method, device, program, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To retrieve all of a desired content section(scene) even when there is any difference on a style(expression) between a character string to be input by a user and a character string included in the subtitle data of a content. <P>SOLUTION: A content retrieval/reproduction device 10 includes a content DB11 for storing the content of a received broadcast signal, a title data decoder part 13 for decoding the title data of content to extract a title text, and for executing the morphemic analysis of the title text, a dictionary DB 12 for morphemic analysis for storing the invariant vocabulary information of conjugated words with self-contained words, an input part 14 for inputting a retrieval keyword, a keyword extracting part 15 for acquiring the invariant vocabulary information corresponding to each conjugated words from the dictionary DB12 when the input retrieval keyword includes the conjugated words with self-contained words and a retrieval part 16 for retrieving the title text including a matched character string, based on the acquired invariant vocabulary information, and for detecting the pertinent content section(scene), based on the retrieval result. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a content search / playback method, a device, a program, and a recording medium on which the program is recorded, for searching and playing back a desired part from a recording medium that receives and records a digital broadcast signal such as a television device. .

  Conventionally, fast-forwarding has been performed in order to find a scene that one wants to see from a recorded video. However, as the capacity of a storage medium increases, searching for a desired part by such fast-forward playback from all recorded programs is time consuming and laborious.

  For example, the document processing apparatus described in Patent Document 1 has a dictionary with word stems and types of utilization, utilization forms and utilization endings, and uses utilization forms for specified phrases in the document. It can be converted to its synonyms and antonyms while retaining them.

  In addition, the electronic dictionary search device described in Patent Document 2, for example, when translating from a Japanese sentence to an English sentence, performs a morphological analysis on a character string input by the user, and corrects a word to be used as a basic form. It is obtained as a headline and searched for a dictionary.

In addition, the video information search / playback method described in Patent Document 3 makes it possible to create a record list in which a subtitle character string and a time code are combined, and to play back from the scene using dialogue as a search key. It is.
Japanese Patent No. 2898983 Japanese Patent No. 3377742 JP-A-11-184867

  The invention described in the above-mentioned Patent Document 1 is intended to solve a search omission by providing a dictionary with all the usage forms and endings of the usage words. There are various tone such as tone, man tone, and woman tone, and it is difficult to have all the endings in the dictionary.

  In the invention described in Patent Document 2, the dictionary search is surely performed by correcting the utilization word to the basic form. However, when the search target is subtitle text, the matching is performed by correcting to the basic form. It may become impossible.

  The invention described in Patent Document 3 detects a portion where the character string input by the user matches the character string of the caption, and the character string style (phrase) input by the user and the caption There is a problem that it is not detected if it does not match the character string style.

  Subtitle data transmitted superimposed on a television broadcast signal is basically a colloquial style (spoken language) spoken by the performer of the content (program) or a narrator. Including various styles such as tone. On the other hand, it is presumed that the request sentence that the user inputs for the search is written in tone when characters are input, and in spoken language when voice is input. As described above, when various styles (phrases) appear, when searching for a user-desired content portion, there is a style mismatch between the character string input by the user and the character string included in the caption data of the content. And may cause a search omission.

  The present invention has been made in view of the above circumstances, and is desirable even if there is a difference in style (phrase) between the character string input by the user and the character string included in the caption data of the content. It is an object of the present invention to make it possible to search the content part without any omission, and to reproduce the content part from a position with a certain degree of separation when reproducing from the content part including the searched character string.

  In order to solve the above-mentioned problem, the first technical means of the present invention obtains subtitle data to be transmitted while being superimposed on a broadcast signal, and collates the subtitle data with a search keyword input by a user. In the content search / playback method for searching / playing back a user-desired content portion from the content of the broadcast signal, the dictionary database for text morphological analysis stores invariant vocabulary information of words used in independent words When the search keyword input by the user is an independent word and includes a usage word, invariant vocabulary information corresponding to each usage word is acquired from the dictionary database, and includes a matching character string based on the acquired invariant vocabulary information. This is characterized in that subtitle data is searched and a corresponding content portion is detected based on the search result. Thereby, even if vocabulary of various styles (phrase) appears in the subtitle text in the content, it is possible to search for a desired content portion without omission.

  A second technical means is the first technical means, wherein the invariant vocabulary information is vocabulary information of an invariant part in any one or a plurality of inflection forms of a verb, an adjective, and an adjective verb that are inflected words in independent words. It is characterized by being.

  According to a third technical means, in the first technical means, the subtitle data is composed of at least one subtitle sentence data unit which is a display unit on the screen, and the subtitle sentence data unit including the searched character string is included. The content portion corresponding to the unit is reproduced. Thereby, it becomes possible to reproduce from a position with a good separation to some extent, and the user can view a desired portion from an appropriate position.

  In the fourth technical means, in the third technical means, when reproducing a plurality of content parts in a single content or a plurality of content parts extending over a plurality of contents, a subtitle sentence data unit is provided for each content. As described above, the reproduction is sequentially performed by forward or backward feeding in time series order.

  In the fifth technical means, in the third technical means, when reproducing a plurality of content parts in a single content or a plurality of content parts extending over a plurality of contents, a subtitle sentence data unit is provided for each content. As described above, according to the operation of the user, it is possible to perform zapping in forward or backward in order of time series.

  According to the sixth technical means, in the third technical means, when reproducing a plurality of content parts in a single content or a plurality of content parts extending over a plurality of contents, the content parts are classified according to a predetermined condition. In addition, for each class, subtitle sentence data units are used as a unit, and playback is performed in order of forward or backward in chronological order.

  According to the third technical means, in the third technical means, when reproducing a plurality of content parts in a single content or a plurality of content parts extending over a plurality of contents, the content parts are classified according to a predetermined condition. In each class, the subtitle sentence data unit is used as a unit, and zapping can be performed in forward or backward in chronological order according to the user's operation.

  The eighth technical means obtains subtitle data to be transmitted while being superimposed on the broadcast signal, and collates the subtitle data with a search keyword input by the user, thereby determining the user-desired content from the content of the broadcast signal. In a content search / playback apparatus for searching / playing back a content part, a dictionary database for morphological analysis storing invariant vocabulary information of a use word as an independent word, input means for inputting a search keyword, and the input search keyword In the case where independent words include inflected words, means for acquiring invariant vocabulary information corresponding to each inflected word from the dictionary database, and searching for subtitle data including a matching character string based on the acquired invariant vocabulary information, Searching means for detecting a corresponding content portion based on the search result is provided.

  A ninth technical means is the eighth technical means, wherein the invariant vocabulary information is vocabulary information of an invariant part in any one or a plurality of inflection forms of a verb, an adjective, and an adjective verb that are inflected words in independent words. It is characterized by being.

  According to a tenth technical means, in the eighth technical means, the subtitle data is composed of at least one subtitle sentence data unit which is a display unit on the screen, and includes a subtitle sentence including a character string searched by the search means. The present invention is characterized in that a reproduction means for reproducing a corresponding content portion in units of data units is provided.

  According to an eleventh technical means, in the tenth technical means, the reproduction means reproduces a plurality of content parts in a single content or a plurality of content parts extending over a plurality of contents by subtitles for each content. A sentence data unit is used as a unit, and playback is performed sequentially in forward or reverse order in time series.

  According to a twelfth technical means, in the tenth technical means, the reproduction means reproduces a plurality of content parts within a single content or a plurality of content parts extending over a plurality of contents, with subtitles for each content. It is characterized in that zapping can be performed in forward or backward in chronological order according to a user's operation in units of sentence data units.

  According to a thirteenth technical means, in the tenth technical means, when the reproducing means reproduces a plurality of content parts in a single content or a plurality of content parts extending over a plurality of contents, the content part is predetermined. According to the above conditions, classification is performed, and for each class, reproduction is performed in order of forward or reverse in time series in units of caption text data units.

  According to a fourteenth technical means, in the tenth technical means, when the reproducing means reproduces a plurality of content parts in a single content or a plurality of content parts extending over a plurality of contents, the content part is predetermined. According to the above conditions, classification is performed, and for each class, zapping can be performed in forward or backward in chronological order according to the user's operation in units of caption text data units.

  A fifteenth technical means is a program for executing the content search / playback method described in any one of the first to seventh technical means.

  The sixteenth technical means is a computer-readable recording medium recording the program described in the fifteenth technical means.

  According to the present invention, subtitle data transmitted by being superimposed on a broadcast signal is obtained, and the subtitle data and a search keyword input by the user are collated, so that a content portion desired by the user can be obtained from the content of the broadcast signal. By using invariant vocabulary information for the search, even if there is a mismatch between the request text entered by the user and the character string included in the subtitles of the content, the desired content portion is leaked. Can be extracted.

  Further, by using the caption text data unit which is a display unit on the screen, it is possible to reproduce from a position with a certain degree of separation, so that the user can view a desired portion from an appropriate position.

  In addition, when a plurality of candidates are detected, zapping can be performed in units of caption text data units, so that a desired portion can be found more quickly.

  The present invention retrieves subtitle data that is superimposed on a broadcast signal and transmits the subtitle data to a search keyword input by the user, thereby searching for a content portion desired by the user from the content of the broadcast signal.・ In the playback method, invariant vocabulary information of words used in independent words is stored in the dictionary for text morphological analysis, and each keyword is used when the search keyword input by the user contains words in words. Invariant vocabulary information corresponding to a word is acquired from a dictionary, subtitle data including a matching character string is searched based on the invariant vocabulary information, and the corresponding content portion can be detected based on the search result. . The desired content part refers to a scene or scene included in the content.

  The inflected words in the above independent words are verbs, adjectives, and adjective verbs, and the inflected words in the independent words include nouns, pronouns, conjunctions, adverbs, conjunctions, and moving verbs. In general, the basic part of the beginning of the word that does not change is called the word stem, but the invariant vocabulary information in the present invention is based on this word stem, and various styles such as a complex tone, a modified tone, a male tone, and a woman tone ( This refers to the unchanging part that does not change in the wording. Note that the ending part that changes is called an inflection ending, and the inflection word basically consists of “stem + inflection ending”.

  In addition, the subtitle text data unit herein refers to a unit of display on the screen of subtitle data transmitted while being superimposed on a broadcast signal. In a digital broadcast system, a broadcast station stores subtitle data in a PES (Packetized Elementary Stream) separate from video and sound information, and multiplexes and sends the subtitle data to a transport stream. Data is separated from the transport stream and analyzed, and subtitles are displayed on the screen. In the present invention, subtitle data is extracted from the subtitle PES data together with the display start time on the screen, or together with the display start time and display end time, and used.

(First embodiment)
FIG. 1 is a block diagram showing a configuration example of a content search / playback apparatus according to a first embodiment of the present invention. In the figure, 10 is a content search / playback apparatus, and the content search / playback apparatus 10 is a content DB 11, dictionary DB 12, caption data decoder unit 13, input unit 14, keyword extraction unit 15, search unit 16, and playback unit 17 are provided. FIG. 1 shows a configuration example of only basic circuit blocks necessary for explaining the content search / playback apparatus 10 according to the present invention.

  In this embodiment, when a request sentence suggesting a part (scene, scene) that the user wants to see is input, a part including a character string included in the request sentence is searched from subtitle data of the content (such as a TV program). The case where it is presented to the user will be described.

  In FIG. 1, a content DB 11 is means for receiving a digital broadcast signal and storing the content (such as a television program). The dictionary DB 12 is a dictionary for morphological analysis, and includes independent word non-use words such as nouns and independent word use words such as verbs, adjectives and adjective verbs, and invariant vocabulary information is also stored for the use words.

  The subtitle data decoder unit 13 decodes subtitle data (for example, closed caption data) multiplexed on the digital broadcast signal, and displays the start time of the subtitle text and each subtitle sentence data unit that is a unit on the screen display. Alternatively, it is means for extracting the display start time and display end time (both are referred to as time code T). Further, the extracted subtitle text is subjected to morphological analysis with reference to the dictionary DB 12. The input unit 14 is a means for the user to input a request sentence.

  The keyword extraction unit 15 performs morphological analysis on the request sentence input from the input unit 14 with reference to the dictionary DB 12, extracts independent words such as nouns, adjectives, adjective verbs, and verbs, and further extracts the extracted character string. However, in the case of an inflection word such as a verb, invariant vocabulary information is acquired with reference to the dictionary DB 12, and these are used as search keywords (hereinafter referred to as search KW).

  The search unit 16 is means for searching the subtitle text stored in the subtitle data decoder unit 13 based on the search KW and obtaining the time code T of the subtitle sentence data unit including the character string. The playback unit 17 is a unit that plays back content from the position of the time code T acquired by the search unit 16.

  Next, the dictionary DB 12 will be described in detail. The dictionary DB 12 performs morphological analysis on the caption data of the content and the request sentence input by the user, and further acquires the invariant vocabulary information when the requested sentence input by the user includes a use word as an independent word. Used. The invariant vocabulary information used here is different from a commonly used stem. Hereinafter, invariant vocabulary information registered in the dictionary DB 12 in the present invention will be described.

For example, when the user inputs “bake cheesecake” as a request sentence, this character string is divided into “cheesecake / bake” by morphological analysis using the dictionary DB 12. The dictionary DB 12 includes, for example, the following information for each word.
Cheese cake (noun)
(Case particle)
Bake (verb, five-stage use, final form)

On the other hand, in the subtitle data included in the content, the following phrases can be considered as phrases related to “burn”.
Bake this in the oven.
What should be noted when baking
Bake on high heat.
Let's bake it .
You can bake it as it is.
It is unlikely that it will burn out.
Once baked , this time,
I bake this whole.
The other day, I decided to bake a cake with my child.
Don't bake immediately, first let it sleep for 1 hour.

There are various expressions as mentioned above, but “baking” is a five-stage use, so it is known to use it as follows.
yak-a-nai / -a-zu (form: negative form)
yak-i-masu / ya (k) -i-ta (continuous use type)
yak-u (end type)
yak-u-toki (continuous form: followed by a noun)
yak-e-ba (assumed form)
yak-e (instruction type)
yak-o-o (oriented form: intention)

  The part common to these is “yak−”. This part is what is generally called a word stem. Therefore, this yak- can be used to search all vocabulary related to “baking” in the subtitle text without omission. However, when collation is performed by text matching, collation is performed only by “burning”. For example, words such as “burnt” and “burning” are also matched. Therefore, “a”, “i”, “u”, “e”, “o” corresponding to the use endings are added to “yak-”, and “baked”, “baked”, “baked”, “burned”, and “baked” are registered in the dictionary DB 12 as invariant vocabulary information. Keep it.

  Furthermore, in the case of a five-stage verb, various stool conversions occur due to the consonant at the end of yak- (in this case k). For example, when ta (ta) follows yak-, it is turned into stool and becomes yai- (baked). Therefore, the vocabulary (in this case, “baked”) taking into account such stool conversion is also registered in the dictionary DB 12 as invariant vocabulary information. In addition to sound stool conversion, there are sound stool (e.g. "bought") and sound-repellent stool (e.g. "read"). In this way, if the word is a self-supporting word and the fifth word is used, the characters that are generally called stems (yak for “baked”) plus a, i, u, e, o A string and a character string in consideration of stool conversion are registered in the dictionary DB 12 as invariant vocabulary information. That is, as the invariant vocabulary information of the verb “baked”, “baked”, “baked”, “baked”, “burned”, “baked”, and “baked” are registered in the dictionary DB 12. If these six types are registered, it is possible to search all of the various phrases mentioned above without omission.

In addition, the usage forms other than the 5-stage utilization are as follows. For example, in the case of “Pleasant”
kob-i-nai
kob-i-rimasu / kab-i-ta
kob-i-ru
kob-i-rutoki
kob-i-reba
kob-i-ro
kob-i-yo

  In this case, kobi- (“apore”), which is a common part, is registered in the dictionary DB 12 as invariant vocabulary information. In addition, in the case of the upper one use, the stool conversion does not occur. Therefore, it is only necessary to register “aporie” as the invariant vocabulary information of the verb “aporie”.

In addition, in the case of “stir fry”, for example,
itam-e-nai
itam-e-masu / itam-e-ta
itam-e-ru
itam-e-rutoki
itam-e-reba
itam-e-ro
itam-e-yoo

  In this case, “item-” (“fried”), which is a common part, is registered in the dictionary DB 12 as invariant vocabulary information. In the case of the lower one, the stool conversion does not occur, so it is only necessary to register “stir fry” as the invariant vocabulary information of the verb “stir fry”.

Further, in the case of a ka-variator, this ka-variator is only “coming”. However, in this case, since the whole word changes, there is no so-called word stem.
ko-nai / -zu
ko-saseru
ki-masu / -ta
ku-ru
kuru-toki
kure-ba
ko-i
ko-yoo

Therefore, in this case, all utilizations (underlined parts) shown below are registered in the dictionary DB 12 as invariant vocabulary information.
Anyone come have a (child), to not a come (this), to come (this), it is very has been found previously to come (this),
Come (this) Re that day when, come (-out) or be, I was coming (-out), Mireba Te came (-out), will Ru come (Ku), come (Ku) Ru day, come ( Ku) Re Bawakaru, have come (this), I thought intends good come (this), come (-out) was Chi Ya',

Also, in the case of the sa variable, for example, “to heat”,
ness-i-nai / -e-zu / -a-reru
ness-i-masu / -i-ta
ness-u-ru
ness-u-rutoki
ness-u-reba
ness-i-ro / -e-yo
ness-i-yo

  In this case, a, i, u, and e that can appear subsequently are added to the common part ness-, and “heat”, “heat”, “heat”, and “heat” are invariant vocabulary information. Is registered in the dictionary DB 12 as

  The search unit 16 collates and searches the caption data stored in the caption data decoder unit 13 based on the search KW (independent word) stored in the keyword extraction unit 15. At this time, if the search KW includes an inflected word as an independent word, a character string search is performed using the invariant vocabulary information registered in the dictionary DB 12 for the collation. That is, when the user inputs “bake cheesecake” as a request sentence, this request sentence includes the word “baked” as an independent word as a search KW. Character string search is performed for all registered invariant vocabulary information “baked”, “baked”, “baked”, “burned”, “baked”, and “baked”. As a result, even if subtitle texts with various expressions such as “I baked cheesecake for the first time last week” and “We can bake cheesecake for the first time” appear, search for the desired content part without omission. can do.

  Next, a specific example of the hardware configuration of the content search / playback apparatus 10 will be described. FIG. 2 is a block diagram showing a specific system configuration example of the content search / playback apparatus 10 shown in FIG. 1. The content search / playback apparatus 10 includes a CPU 1, a work memory 2, an input unit 3, a database 4, and a display. Unit 5, sound output unit 6, external storage drive 7, and reception unit 8. The CPU 1 controls the entire system of the content search / playback apparatus 10 according to a built-in system program, and executes various programs such as a search processing program corresponding to the search unit 16 shown in FIG. Further, content such as a television program received by the receiving unit 8 is stored in the database 4 corresponding to the content DB 11 shown in FIG. The CPU 1 processes the content stored in the database 4 and creates a processing result on the work memory 2.

  The work memory 2 is a so-called working memory that temporarily stores data related to the above-described search KW, data used for calculation, calculation results, and the like, and is used in the search program executed by the CPU 1 and the search program. Stored data, program processing results, and the like. Note that some of the search programs that are unlikely to be changed may be stored in a ROM (not shown) as in the case of the system program.

  In addition, a part of the RAM used as the work memory 2 is backed up by power, or is configured by a non-volatile memory such as an EEPROM (Electrically Erasable Programmable ROM) or a flash memory, and the power is off. Even in this case, it is possible to retain the stored contents as they are. In this nonvolatile storage area, various data including processing KW, processing data, programs, and the like that need to retain stored contents even when the power is turned off are stored.

  The input unit 3 corresponds to the input unit 14 shown in FIG. 1, and includes a remote controller key, a microphone, a camera, a sensor, and the like for inputting a request text by the user. The camera or sensor may be provided with a character recognition function so that characters such as newspapers and magazines can be input. The character string input from the input unit 3 is processed by the CPU 1 and stored in the work memory 2 as a search KW.

  The display unit 5 and the audio output unit 6 correspond to the reproduction unit 17 shown in FIG. 1 and extract a video signal and an audio signal for TV screen display from the broadcast signal received by the receiving unit 8. While displaying, the various information stored in the database 4 and the arithmetic processing result of CPU1 are output. The display unit 5 includes a liquid crystal display or a plasma display, and the sound output unit 6 includes a speaker, headphones, or the like. The external storage drive 7 is a reading unit of an external storage medium (recording medium) 9 in which various broadcast reception programs including programs executed by the CPU 1 are recorded.

  The external storage medium (recording medium) 9 includes recording media such as various memory cards, hard disks, and DVD-ROMs. Here, the external storage medium (recording medium) 9 stores a content search / playback processing program or the like in an embodiment to be described later, and the external storage medium (recording medium) 9 is attached to the external storage drive 7. Various information stored in the memory is read and stored in the work memory 2 or the like.

  Hereinafter, an operation example of the content search / playback apparatus 10 will be described with reference to FIG. As described above, according to the present invention, a portion (scene) that a user desires to view content (such as a TV program) received as a broadcast signal using subtitle data included in the broadcast signal and transmitted. , Scene) and is intended to be presented to the user. Hereinafter, a case where a desired portion is searched based on a request sentence input by the user using a key on the remote controller and presented to the user will be described as an example.

  First, a basic operation example of each block of the content search / reproduction device 10 shown in FIG. 1 will be described. The content DB 11 stores received digital content (such as broadcast programs). As described above, the data stored here includes subtitle data multiplexed with the received broadcast signal in addition to the video / sound data. The dictionary DB 12 is a dictionary for morphological analysis, and includes invariant vocabulary information of words used as independent words. The caption data decoder unit 13 decodes the caption data (closed caption data and the like) included in the content DB 11, and displays the caption text and the display start time or the display start time of each caption sentence data unit that is a unit on the screen display. -Take out the display end time (time code T). Further, the extracted subtitle text is subjected to morphological analysis with reference to the dictionary DB 12. The input unit 14 inputs a request sentence (text) related to a portion (scene, scene) that the user wants to see with a key on the remote controller or the like. The input request sentence is output to the keyword extraction unit 15.

  The keyword extraction unit 15 performs morphological analysis on the user's request sentence input from the input unit 14 with reference to the dictionary DB 12, and extracts independent words such as nouns, adjectives, and verbs. Further, for inflected words, invariant vocabulary information is acquired with reference to the dictionary DB 12. For example, if the request sentence is “how to pickle kimchi”, the morphological analysis is as follows: / kimchi (noun) / no (participant) / pick (verb: lower one step utilization) / method (adjunct) / First, the noun “Kimchi” which is an independent word and the verb “Kizuke” are extracted, and “zuke” which is invariant vocabulary information of the verb “Kizuke” is acquired as a search KW. Here, since the verb “pickle” is used further below, the only invariant vocabulary information is “pickle”. The extracted search KW “kimchi” and “duke” are output to the search unit 16.

  The search unit 16 searches the caption data stored in the caption data decoder unit 13 based on the search KW extracted by the keyword extraction unit 15. That is, it is searched whether or not a character string matching “Kimchi” and “Zuke” exists in the caption data of the content. When a matching character string is detected, the time code T of the caption text data unit including the character string is acquired, and the time code T is output to the playback unit 17.

  The reproduction unit 17 reproduces and outputs the content stored in the content DB 11 from the position of the time code T extracted by the search unit 16. At this time, the caption data may or may not be reproduced. Further, if necessary, the user can return to a position before the position corresponding to the time code T by a remote control operation or the like. When the time code T includes the display end time, the reproduction is stopped at the time. When the time code T does not include the display end time, the content is played to the end unless the user stops it with a remote controller or the like.

  Here, the caption text data unit will be described with reference to the schematic diagram shown in FIG. The broadcast reception signal input to the content DB 11 is composed of video data, sound data (sound and sound), caption data, and the like. Here, the caption text data unit is a display unit on the screen and is transmitted together with the display start time (time code T) = 101, 102,. For example, in FIG. 3, the caption data “Continue to make Chinese cabbage kimchi” is displayed from time T = 102 to time T = 103, and then the caption is displayed from time T = 103 to time T = 104. The data “First, salt the Chinese cabbage salted the previous day” is displayed. That is, the display start time of the subsequent caption text data unit is the display end time of the previous caption text data unit.

  The search unit 16 searches subtitle data stored in the subtitle data decoder unit 13 based on the search KW extracted by the keyword extraction unit 15, and when a character string that matches the search KW is detected, The time code T of the caption text data unit including the character string is acquired, and the time code T is output to the playback unit 17. As a result, it is possible to reproduce from a position with a good separation.

  Next, an operation example of the search unit 16 will be described in detail. The search unit 16 searches the subtitle data in the subtitle data decoder unit 13 and first checks whether there is content including the search KW. If there are a plurality of search KWs, it is checked whether there is content including all the search KWs. (However, not all search KWs need to be included in one caption text data unit). If the content exists, the time code T of the subtitle sentence data unit including the character string that matches each search KW in the content is acquired, and the time code T is output to the playback unit 17.

  For example, if the search KW is “Kimchi” and “Kizuke”, the subtitle sentence data unit detected in FIG. 3 is “Next, we will make a Chinese cabbage kimchi” and “Now, the delicious Chinese cabbage kimchi is soaked. The time codes T “102-103” and “106-107” are acquired and output to the reproduction unit 17. The time code T may be only the display start time.

  Based on the time code T, the playback unit 17 plays back the content in the content DB 11 in units of caption text data units. That is, the reproduction unit 17 reproduces and outputs a portion corresponding to “Continue to make Chinese cabbage kimchi” and “Now, delicious Chinese cabbage kimchi is pickled.”

  As described above, the playback unit 17 plays back the content in the content DB 11 in units of subtitle sentence data units based on the time code T acquired by the search unit 16. FIG. 4 is a diagram illustrating an example of a screen when the content search / playback apparatus 10 performs playback processing. On the screen, a detected content name (a cooking program in this example) 21, a caption text data unit group 22 including a plurality of caption text data units corresponding to the content of the content name 21, and a playback video 23 are displayed. Displayed at the top. The subtitle sentence data unit group 22 is a subtitle sentence data unit detected by the search unit 16 as a center and the subtitle sentence data units before and after the subtitle sentence data unit are displayed as thumbnails. T in FIG. 4 represents the time code T of the detected caption text data unit. Of course, only the portion corresponding to the time code T may be displayed as a thumbnail. At this time, subtitles may or may not be displayed. Further, the subtitle and the video may be displayed in separate windows. Also, there may be no thumbnail.

  In FIG. 4, when playing back the content, if the user selects an arbitrary thumbnail by remote control operation or the like, it may be started from the selected position T (in this case, stopped in units of subtitle sentence data units). Alternatively, playback may be continued until the end of the content until the user stops by remote control operation or the like), and automatic frame-by-frame playback in units of subtitle sentence data units in time-sequential order of time code T You may make it do.

  During playback, the user may return to the position of the time code T by operating the remote control or the like, or during playback, the user can stop playback by operating the remote control or the like. You may do it. Further, the reproduction start position may be made coincident with the time code T, or may be reproduced from a certain fixed time (for example, 1 second).

  Further, the playback start position may be adjusted using video information or sound information. In other words, by using video information and sound information in addition to the detected time code T of the caption text data unit 22, it is possible to determine a more appropriate position for starting viewing. As sound information, for example, it is possible to determine a more appropriate playback start position by using the presence / absence or change of background sound, change of noise level, change of speaker using voice recognition, change of silence state, etc. . As video information, for example, a more appropriate reproduction start position can be determined by using a motion vector, a luminance level, a change amount of a chromaticity histogram, and the like. Therefore, an optimal playback start position may be determined from the viewpoint of video information and / or sound information, including a caption text data unit.

  Next, an operation example of the search unit 16 in the content search / playback apparatus 10 shown in FIG. 1 will be further described. First, when the content search / playback apparatus 10 is activated, the broadcast reception signal is stored in the content DB 11, and further, caption data (such as closed caption data) included in the broadcast reception signal is transmitted to the caption data decoder unit 13. Output. The subtitle data decoder 13 decodes the input subtitle data into a text format, and extracts the subtitle text and the time code T of the subtitle sentence data unit. Further, the extracted subtitle text is subjected to morphological analysis with reference to the dictionary DB 12. The processing in the caption data decoder unit 13 may be executed sequentially every time new content is received in the content DB 11 or may be executed at certain time intervals.

  FIG. 5 is a flowchart for explaining an operation example in the search unit 16 of the content search / playback apparatus 10 shown in FIG. Here, an example of a search operation when the user inputs a request sentence “how to pickle kimchi” will be described. First, when the input unit 14 receives a character string input by the user (step S1), the keyword extraction unit 15 performs a morphological analysis of the request sentence using the dictionary DB 12 (step S2), and uses an independent word non-use word such as a noun. And invariant vocabulary information of the independent word utilization words (verbs, adjectives, adjective verbs) are acquired and output to the search unit 16 as a search KW (step S3). That is, in this example, “kimchi” and “duke” are output to the search unit 16 as the search KW.

  Here, when the user's request sentence is, for example, “I want to know how to pickle kimchi”, in this request sentence, “I want to know” is irrelevant to the content of the content, so it is not a search KW. In this way, processing that does not include expressions representing user requests such as “I want to know” and “I want to see” in the search KW.

  The search unit 16 collates the extracted search KW with the text in the subtitle data analyzed and extracted by the subtitle data decoder unit 13 to determine whether or not there is content including all the search KWs in the subtitle data. Is discriminated (step S4). That is, in this example, it is determined whether or not there is content including both of the search KW “Kimchi” and “Kizuke”.

  In step S4, when content including all of the search KWs is detected (in the case of YES), the subtitle sentence data unit including each of the search KWs is searched, the time code T is acquired, and output to the playback unit 17 (Step S5). The reproduction unit 17 reproduces and outputs the portion corresponding to the time code T from the content stored in the content DB 11 (step S6), and ends the process.

  On the other hand, if it is determined in step S4 that there is no content including all of the search KWs (in the case of NO), the playback unit 17 returns to the request text accepting state in step S1 without playing back the content, and the user Prompts for a request sentence containing another vocabulary.

  Note that the matching process between the search KW and the caption data in step S4 is a so-called text search process, and it is possible to use a search technique that is used in a homepage search or the like on the Internet. That is, the vocabulary (word, phrase, etc.) of subtitle data that completely matches the search KW may be detected, or a semantic (or morphological) similarity is obtained and determined in advance. It is also possible to detect a vocabulary having a similarity greater than a certain threshold.

(Second embodiment)
In the present embodiment, an example of a reproduction method when a plurality of parts (scenes, scenes) are detected will be described. When a plurality of portions (scenes, scenes) are detected in step S4 of the flow shown in FIG. 5, the search unit 16 acquires a plurality of time codes T. In this case, the following two cases can be considered.
(1) Multiple parts (scenes, scenes) in one content
(2) Multiple parts (scenes, scenes) that span multiple contents

  When a plurality of parts (scenes, scenes) are candidates as described above, the following reproduction method can be considered. The display method on the screen is not limited to the example shown below. For example, when two contents are detected, as shown in FIG. 6, two detected content names (in this example, (Cooking program, variety) 31, a caption text data unit group 32 composed of a plurality of caption text data units corresponding to each content of the content name 31, and a playback video 33 are displayed at the top of each caption text data unit group 32. The The subtitle sentence data unit group 32 is a subtitle sentence data unit before and after the subtitle sentence data unit detected by the search unit 16 displayed as a thumbnail for each content. In the example shown in FIG. 6, thumbnails before and after the detected caption text data unit are also displayed to make the portion corresponding to the time code T stand out by coloring or the like, but only the portion corresponding to the time code T is displayed. May be displayed as thumbnails. At this time, subtitles may or may not be displayed. In this state, the user selects a desired content by operating a remote controller or the like, and the subtitle sentence data unit is automatically framed and presented for each selected content in chronological order of the time code T. Alternatively, the subtitle sentence data units may be zapped by the user himself / herself through remote control operation or the like in the time-series order of the time code T.

  Further, for example, as shown in FIG. 7, the extracted caption text data unit (that is, the content part or the entire content including the caption sentence data unit) is classified into classes for each unit such as a genre and a topic (topic). It is presented separately (class 41 in the figure), and for each class selected by the user (in this example, one of music, variety, or drama), each content is automatically or by remote control by the user. The subtitle sentence data units may be zapped in time sequence of the time code T. In particular, when a large number of candidates are extracted, it takes time to view in chronological order, but by classifying in this way, the user can find a desired part (scene, scene) faster. It becomes possible.

  The order of contents to be presented to users in each class may be presented in chronological order such as production date or broadcast date, or includes more (longer) extracted subtitle sentence data units. The content may be presented with a higher priority. Alternatively, the viewer may present it separately according to whether it has been viewed or not viewed.

  As described above, each embodiment has been described centering on each function in the content search / playback apparatus of the present invention. However, as described above, the present invention is also a content search as a method for executing each step in the content search / playback apparatus. -It can take the form as a reproduction method. In addition, a form as a program for causing a computer to execute the content search / playback method, and a form as a recording medium on which the program is recorded are possible.

  An embodiment of a recording medium storing a program and data for realizing a content search / playback function according to the present invention will be described. Specific examples of the recording medium include CD-ROM (-R / -RW), magneto-optical disk, DVD-ROM (-R / -RW), FD, HD, flash memory, memory card, memory stick, Various other ROMs, RAMs, and the like can be assumed. A program for causing a computer to execute the functions as the content search / playback apparatus according to each embodiment of the present invention described above on these recording media and realizing the content search / playback functions is provided. By recording and distributing, it is easy to realize the function. Then, the recording medium as described above is mounted on an information processing apparatus such as a computer and the program is read by the information processing apparatus, or the program is stored in a storage medium provided in the information processing apparatus. By reading, it is possible to execute the content search / playback function according to the present invention.

It is a block diagram which shows the structural example of the content search / reproduction apparatus based on 1st Example of this invention. FIG. 2 is a block diagram illustrating a specific system configuration example of the content search / playback apparatus illustrated in FIG. 1. It is the figure which showed the caption text data unit typically. It is a figure which shows the example of a screen when reproducing | regenerating processing in a content search / reproducing apparatus. It is a flowchart for demonstrating the operation example in the search part of the content search / reproducing apparatus shown in FIG. It is a figure which shows the other example of a screen at the time of reproducing | regenerating processing in a content search / reproducing apparatus. It is a figure which shows the other example of a screen at the time of performing a reproduction process in a content search / reproduction device.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... CPU, 2 ... Work memory, 3,14 ... Input part, 4 ... Database, 5 ... Display part, 6 ... Sound output part, 7 ... External storage drive, 8 ... Receiving part, 9 ... External storage medium (recording medium) ) 10 ... content search / playback device, 11 ... content DB, 12 ... dictionary DB, 13 ... subtitle data decoder section, 15 ... keyword extraction section, 16 ... search section, 17 ... playback section, 21, 31 ... content name, 22, 32 ... subtitle sentence data unit group, 23, 33 ... reproduced video, 41 ... class, 101 to 107 ... display start time (time code T).

Claims (16)

  By retrieving subtitle data that is superimposed on the broadcast signal and transmitting it, and collating the subtitle data with a search keyword input by the user, a user-desired content portion is retrieved and reproduced from the content of the broadcast signal. In the content search / playback method, if the dictionary database for morphological analysis of text has invariant vocabulary information of inflected words stored in independent words, and the search keyword input by the user includes inflected words in autonomous words, Invariant vocabulary information corresponding to each utilized word is acquired from the dictionary database, subtitle data including a matching character string is searched based on the acquired invariant vocabulary information, and the corresponding content portion is detected based on the search result A content search / playback method characterized by:   2. The content search / playback method according to claim 1, wherein the invariant vocabulary information is vocabulary information of an invariant part in any one or more of verbs, adjectives, and adjective verbs that are inflected words as independent words. A content search / playback method characterized by being.   The content search / playback method according to claim 1, wherein the subtitle data is composed of at least one subtitle sentence data unit that is a display unit on a screen, and the subtitle sentence data unit including the searched character string is a unit. A content search / playback method characterized in that the corresponding content portion is played back.   4. The content search / playback method according to claim 3, wherein when a plurality of content parts in a single content or a plurality of content parts extending over a plurality of contents are played back, a subtitle sentence data unit is used for each content as a unit. A content search / reproduction method, wherein reproduction is performed sequentially in forward or backward in time series order.   4. The content search / playback method according to claim 3, wherein when a plurality of content parts in a single content or a plurality of content parts extending over a plurality of contents are played back, a subtitle sentence data unit is used for each content as a unit. A content search / playback method characterized in that zapping can be performed in forward or backward in chronological order according to a user's operation.   4. The content search / reproduction method according to claim 3, wherein when reproducing a plurality of content parts in a single content or a plurality of content parts extending over a plurality of contents, the content parts are classified according to a predetermined condition. A content search / playback method, wherein for each class, a subtitle sentence data unit is used as a unit, and playback is performed in order of forward or backward in chronological order.   4. The content search / reproduction method according to claim 3, wherein when reproducing a plurality of content parts in a single content or a plurality of content parts extending over a plurality of contents, the content parts are classified according to a predetermined condition. A content search / playback method characterized in that, for each class, zapping can be performed forward or backward in chronological order according to a user's operation in units of subtitle sentence data units.   By retrieving subtitle data that is superimposed on the broadcast signal and transmitting it, and collating the subtitle data with a search keyword input by the user, a user-desired content portion is retrieved and reproduced from the content of the broadcast signal. In the content search / playback device, a dictionary database for morphological analysis storing invariant vocabulary information of words used in independent words, input means for inputting search keywords, and the input search keywords include words used in words independent In this case, a means for acquiring invariant vocabulary information corresponding to each utilization word from the dictionary database, and subtitle data including a character string that matches based on the acquired invariant vocabulary information are searched, and the corresponding result is obtained based on the search result. A content search / playback apparatus comprising search means for detecting a content portion.   9. The content search / playback apparatus according to claim 8, wherein the invariant vocabulary information is vocabulary information of an invariant part in any one or more of verbs, adjectives, and adjective verbs that are inflected words as independent words. A content search / playback device characterized by being.   9. The content search / playback apparatus according to claim 8, wherein the subtitle data includes at least one subtitle text data unit that is a display unit on a screen, and includes subtitle text data searched by the search means. A content search / playback apparatus comprising playback means for playing back a corresponding content portion in units.   11. The content search / playback apparatus according to claim 10, wherein the playback unit plays back a subtitle sentence for each content when playing back a plurality of content parts in a single content or a plurality of content parts that span a plurality of contents. A content search / playback apparatus, which plays back data in units of data units in time-sequential order and forward or backward.   11. The content search / playback apparatus according to claim 10, wherein the playback unit plays back a subtitle sentence for each content when playing back a plurality of content parts in a single content or a plurality of content parts that span a plurality of contents. A content search / playback apparatus characterized in that zapping can be performed in forward or backward in chronological order according to a user's operation in units of data units.   11. The content search / playback apparatus according to claim 10, wherein when the playback unit plays back a plurality of content parts in a single content or a plurality of content parts spanning a plurality of contents, A content search / reproduction apparatus, wherein classification is performed according to conditions, and reproduction is sequentially performed in forward or backward in time series in units of subtitle sentence data units for each class.   11. The content search / playback apparatus according to claim 10, wherein when the playback unit plays back a plurality of content parts in a single content or a plurality of content parts spanning a plurality of contents, A content search / playback apparatus characterized in that it is classified according to conditions and zapping can be performed in forward or backward in chronological order according to a user's operation in units of subtitle sentence data units for each class.   A program for executing the content search / playback method according to any one of claims 1 to 7.   The computer-readable recording medium which recorded the program of Claim 15.

Images