CN112269900A - Method for rapidly retrieving collected video - Google Patents
Method for rapidly retrieving collected video Download PDFInfo
- Publication number
- CN112269900A CN112269900A CN202011094314.4A CN202011094314A CN112269900A CN 112269900 A CN112269900 A CN 112269900A CN 202011094314 A CN202011094314 A CN 202011094314A CN 112269900 A CN112269900 A CN 112269900A
- Authority
- CN
- China
- Prior art keywords
- video
- synonym
- retrieval
- acquiring
- webpage
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/237—Lexical tools
- G06F40/247—Thesauruses; Synonyms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/73—Querying
- G06F16/735—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/10—Text processing
- G06F40/103—Formatting, i.e. changing of presentation of documents
- G06F40/117—Tagging; Marking up; Designating a block; Setting of attributes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
Abstract
The invention provides a method for rapidly retrieving and collecting videos. Acquiring a search word of a user, and determining a synonym set of the search word; constructing a constraint model of the retrieval video according to the synonym set, retrieving and acquiring the retrieval video set corresponding to the synonym set according to the constraint model; and screening videos in the retrieval video set through preset characteristic parameters according to the characteristics of the synonym set in the video set corresponding to the content to obtain a target video set. The invention discloses a method for quickly retrieving and acquiring videos, which is mainly used for acquiring a large number of similar videos by a user; the synonyms of the videos are extracted, so that the videos related to the search terms can be obtained to the maximum extent.
Description
Technical Field
The invention relates to the technical field of video retrieval, in particular to a method for rapidly retrieving and acquiring videos.
Background
At present, the industry of short videos is prosperous because the production of short videos requires a large amount of materials. However, each company has also completed the retrieval and collection of video data through ordinary video downloading and capital-consuming purchases, which has led to a substantial increase in cost. Moreover, the video acquisition efficiency is extremely low, and the user can hardly acquire the own intention video.
Disclosure of Invention
The invention provides a method for rapidly retrieving and collecting videos. The method is used for solving the problem of difficult video acquisition.
A method for rapidly retrieving a captured video, comprising:
acquiring a search word of a user, and determining a synonym set of the search word;
constructing a constraint model of the video according to the synonym set, and retrieving and acquiring a retrieval video set corresponding to the synonym set according to the constraint model;
and acquiring the characteristics of the corresponding content of the synonyms in the synonym set in the retrieval video set, and screening the videos in the retrieval video set through preset characteristic parameters to obtain a target video set.
As an embodiment of the present invention: the obtaining of the search term of the user and the determining of the synonym set of the search term comprise:
acquiring a search term;
obtaining a first synonym set A ═ { a ═ of the search words according to a dictionary1,a2L L ax}；
Acquiring a second synonym set B ═ { B ═ B of the search term according to the webpage label1,b2L L by}；
Acquiring a third synonym set C ═ { C ═ C of the search term according to the text relation1,c2L L cz}；
Performing the same item analysis on the first synonym set A, the second synonym set B and the third synonym set C to obtain a same item set S(A,B,C):
Wherein, the axRepresenting the xth item in the first synonym set A; b isyRepresents the y item in the second synonym set B; the above-mentionedczRepresenting the z item in the first synonym set C; 1,2,3L x; 1,2,3L y; 1,2,3, L z;
according to the same item set S(A,B,C)And determining the synonym set.
As an embodiment of the present invention: the constructing of the constraint model of the video according to the synonym set comprises the following steps:
step 1: and the distance D between the synonym set and the search word is as follows:
D＝(G-si)P-1(G-si)；
wherein G represents the search term; the P is-1A covariance matrix of the synonym set, siAn ith synonym representing the set of synonyms;
step 2: constructing a linear regression equation of the synonym set;
wherein xi is the inner product of the synonym set; the above-mentioned
Represents a mean of the set of synonyms;
and step 3: determining a constraint model F of the video according to the linear regression equation of the distance D:
wherein, k isjA jth video representing a set of retrieved videos; the above-mentioned
Means representing a set of retrieved videos; the beta represents an error coefficient of the retrieved video set.
As an embodiment of the present invention: the retrieving and acquiring the retrieval video set corresponding to the synonym set according to the constraint model comprises the following steps:
acquiring a first retrieval video set from any video retrieval website according to the synonym set;
determining a mean value of the first search video set;
importing the parameters of the retrieved videos in the retrieved video set into the constraint model according to the average value to obtain a constraint value;
and acquiring a positive value in the constraint value and a video corresponding to the positive value, and determining the retrieval video collection.
As an embodiment of the present invention: the retrieving and acquiring the retrieved video set corresponding to the synonym set according to the constraint model further comprises:
retrieving videos in any video retrieval website according to the synonym set;
acquiring a link address of the retrieval video;
determining a video webpage corresponding to the link address according to the link address;
deleting the link address when the link address does not have the corresponding video webpage;
triggering videos in the video webpage, responding to the triggering behavior of the video webpage, and acquiring videos of the video webpage to form the retrieval video set when the triggering behavior of the video webpage has a response;
and when the triggering behavior of the video webpage has no response, judging that the video webpage is an invalid webpage.
As an embodiment of the present invention: the retrieving and acquiring the retrieved video set corresponding to the synonym set according to the constraint model further comprises:
acquiring a network transmission protocol of the retrieval video;
judging the acquisition mode of the retrieval video according to the network transmission protocol;
acquiring a corresponding video loading tool according to the acquisition mode of the retrieval video;
and acquiring the retrieval video set according to the loading tool.
As an embodiment of the present invention: the retrieving and acquiring the retrieved video set corresponding to the synonym set according to the constraint model further comprises:
acquiring a video webpage corresponding to the retrieval video;
judging the frequency of abnormal pop-up of the video webpage;
judging whether the video webpage is an abnormal webpage or not according to the times of the abnormal popup;
when the video webpage is an abnormal webpage, deleting the video webpage;
and when the video webpage is a normal webpage, acquiring the video of the video webpage to form a retrieval video set.
As an embodiment of the present invention: the obtaining of the characteristics of the synonyms of the synonym set appearing in the corresponding content of the synonyms in the video set, and screening the videos in the retrieval video set through preset characteristic parameters to obtain a target video set includes:
after the retrieval video set is determined, acquiring the semanteme of each synonym in the synonym set;
judging the characteristics of the video corresponding to the synonyms according to the semantic meaning of each synonym in the synonym set;
and determining a final target video set according to the comparison between the characteristics of the videos corresponding to the synonym set and the parameters of the template video.
As an embodiment of the present invention: the preset characteristic parameters are determined by the following steps:
step S1: acquiring a retrieval video set E ═ E1,e2L L eg}; the g represents the g video; 1,2,3L g;
step S2: determining a feature matrix R of the retrieval video set according to the retrieval video set E:
wherein, r isglThe ith feature representing the ith video;
step S3: calculating a difference coefficient eta of the retrieval video set according to the characteristic matrix R of the retrieval video set:
wherein the em is a characteristic parameter of the template video, and the N is the number of retrieval video sets;
step S4: determining a threshold value epsilon of the target video as em eta according to the difference coefficient eta and the template video:
when the characteristic parameter of the video in the retrieval video set is larger than or equal to the threshold value of the target video, determining that the video in the retrieval video set is the target video;
when retrieving video e in the video setgWhen the characteristic parameter of the search video set is smaller than the threshold value of the target video, deleting the video e in the search video setg。
As an embodiment of the present invention: in step S2, determining a feature matrix of the search video set includes the following steps:
step S10, obtaining synonym semantics according to synonyms corresponding to the videos in the retrieved video set;
step S11: according to the synonym semantics, performing semantic marking in the video of the retrieval video set;
step S12, extracting the frame image of the semantic mark according to the semantic mark and acquiring the characteristics of the frame image;
step S13: and constructing a feature matrix of the retrieval video set according to the features of the frame images.
The invention has the beneficial effects that: the invention discloses a video acquisition method, which is mainly used for acquiring a large amount of similar videos by a user; the video related to the search terms can be obtained to the maximum extent by extracting the synonym set of the video. And (3) constructing a constraint model through synonyms, further screening the videos, and extracting the videos only related to the search terms. During video extraction, judgment can be performed through webpage abnormity judgment, effectiveness judgment of video link addresses and an acquisition mode, so that videos of invalid webpages can be deleted and Trojan viruses can be placed in an acquired video set. And the judgment of the acquisition mode can enable video acquisition personnel to delete videos which cannot be acquired according to the existing video downloading tool or loading tool. And deleting videos with low semantic appearance characteristics of the search words through the semantic appearance characteristics of the search words to obtain an accurate target video collection, so that accurate video acquisition can be performed in the fields of short video production, video directional acquisition and the like, and the situation that a large amount of manpower, material resources and financial resources are consumed for the short video production and the video acquisition is prevented.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and drawings.
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:
fig. 1 is a flowchart of a method for quickly retrieving a captured video according to an embodiment of the present invention.
Detailed Description
The preferred embodiments of the present invention will be described in conjunction with the accompanying drawings, and it will be understood that they are described herein for the purpose of illustration and explanation and not limitation.
Example 1:
as shown in fig. 1, the present invention includes a method for rapidly retrieving a captured video, comprising:
step 100: acquiring a search word of a user, and determining a synonym set of the search word; in the embodiment of the invention, when a user has a retrieval request, for example, the user wants to search for a type A movie, the user can send a video retrieval request through a retrieval word carrying the type A movie, wherein the video retrieval request is identity information for uniquely identifying a certain type of video, such as a movie name and movie features of the type A movie.
Step 101: constructing a constraint model of the retrieval video according to the synonym set, retrieving and acquiring the retrieval video set corresponding to the synonym set according to the constraint model; the constraint model constructed by the synonyms is a screening model, the required video can be accurately screened from the content by taking the linear regression function of the retrieval words and the synonyms as the constraint reference, and after screening, the judgment of abnormity, effective value judgment and acquisition mode of the downloaded webpage of the video is required.
Step 102: and acquiring the characteristics of the synonym set in the video set corresponding to the content, and screening the videos in the retrieval video set through preset characteristic parameters to obtain a target video set. The feature screening is a secondary screening of the target video, because the retrieved video is a video set, inevitably, the videos are interspersed with the related content of the search terms, but not all videos are related content, so the videos need to be deleted, and the problem that the density of the search terms in the retrieved videos is too low can be effectively solved by screening the videos through the features appearing in the meaning of the search terms.
The invention has the beneficial effects that: the invention discloses a video acquisition method, which is mainly used for acquiring a large amount of similar videos by a user; the video related to the search terms can be obtained to the maximum extent by extracting the synonym set of the video. And (4) constructing a constraint model through the synonym set, further screening the videos, and extracting the videos only related to the search terms. During video extraction, judgment can be performed through webpage abnormity judgment, effectiveness judgment of video link addresses and an acquisition mode, so that videos of invalid webpages can be deleted and Trojan viruses can be placed in an acquired video set. And the judgment of the acquisition mode can enable video acquisition personnel to delete videos which cannot be acquired according to the existing video downloading tool or loading tool. And deleting videos with low semantic appearance characteristics of the search words through the semantic appearance characteristics of the search words to obtain an accurate target video collection, so that accurate video acquisition can be performed in the fields of short video production, video directional acquisition and the like, and the situation that a large amount of manpower, material resources and financial resources are consumed for the short video production and the video acquisition is prevented.
Example 2:
as an embodiment of the present invention: the obtaining of the search term of the user and the determining of the synonym set of the search term comprise:
acquiring a search term;
obtaining a first synonym set A ═ { a ═ of the search words according to a dictionary1,a2L L ax}；
Acquiring a second synonym set B ═ { B ═ B of the search term according to the webpage label1,b2L L by}；
Acquiring a third synonym set C ═ { C ═ C of the search term according to the text relation1,c2L L cz}；
Performing the same item analysis on the first synonym set A, the second synonym set B and the third synonym set C to obtain a same item set S(A,B,C):
Wherein, the axRepresenting the xth item in the first synonym set A; b isyRepresents the y item in the second synonym set B; c is mentionedzRepresenting the z item in the first synonym set C; 1,2,3L x; 1,2,3L y; 1,2,3, L z;
according to the same item set S(A,B,C)And determining the synonym set.
The principle of the invention is as follows: in the same acquisition step, the dictionary, the webpage and the text relation are used, and the dictionary comprises a Xinhua dictionary, a knowledge map-based dictionary, a Chinese and foreign language inter-translation dictionary and the like. The web pages include various web page encyclopedias, such as encyclopedia, dog search encyclopedia, and the like. And the text relation means that the meanings of the expressions are the same as the search terms in some texts, and the expressions can be replaced mutually. After all synonyms are included, repeated synonyms are screened out through simple similar screening of the synonyms, and then synonym sets of the same items are obtained.
The invention has the beneficial effects that: the method has wide range of obtaining synonyms, can obtain synonyms to the maximum extent, but the preliminarily obtained synonyms always have deviation, and further obtain repeated synonyms through repeated screening, so that the synonyms do not have deviation in meaning, and further the relevance of videos and search words in the video set can be improved.
Example 3:
as an embodiment of the present invention: the constructing of the constraint model of the retrieval video according to the synonym set comprises the following steps:
step 1: and the distance D between the synonym set and the search word is as follows:
D＝(G-si)P-1(G-si)；
wherein G represents the search term; the P is-1A covariance matrix of the synonym set, siAn ith synonym representing the set of synonyms;
step 2: constructing a linear regression equation of the synonym set;
wherein xi is the inner product of the synonym set; the above-mentioned
Represents a mean of the set of synonyms;
and step 3: determining a constraint model F of the video according to the linear regression equation of the distance D:
wherein, k isjA jth video representing a set of retrieved videos; the above-mentioned
Means representing a set of retrieved videos; the beta represents an error coefficient of the retrieved video set.
The constraint model for constructing the video is constructed by the Mahalanobis distance of the synonym and the search word and the linear regression parameter of the synonym. The mahalanobis distance enables the acquired video and the search terms to have relevance, and the video without the relevance is screened out. The linear regression function can make the obtained video be the video obtained according to the search term. The difference between the retrieved video and the video mean determines that the finally determined video must be a parameter greater than the mean standard, and the product of subtracting the error coefficient and the sum of the video and the video mean must be greater than 1; the error parameter is less than 1. And the product of the error parameter times the mahalanobis distance and the sum of the video and the video mean must be less than the product of the mahalanobis distance times the difference of the video and the video mean. And finally, the value obtained by the constraint function is also a positive value, so that the video set can be obtained by the constraint model.
Example 4:
as an embodiment of the present invention: the retrieving and acquiring the retrieval video set corresponding to the synonym set according to the constraint model comprises the following steps:
acquiring a first retrieval video set from any video retrieval website according to the synonym set;
determining a mean value of the first search video set;
importing the parameters of the retrieved videos in the retrieved video set into the constraint model according to the average value to obtain a constraint value;
and acquiring a positive value in the constraint value and a video corresponding to the positive value, and determining the retrieval video collection.
The principle of the invention is as follows: and after the video set is determined, obtaining a video set constraint value through the mean value, and further obtaining a retrieved video through the constraint value.
The invention has the beneficial effects that: according to the method, the retrieval videos obtained through the constraint model are screened through the constraint value, so that the retrieved videos are all retrieval videos related to the retrieval words, and the accuracy of the obtained videos is guaranteed to the maximum extent.
Example 5:
as an embodiment of the present invention: the retrieving and acquiring the retrieved video set corresponding to the synonym set according to the constraint model further comprises:
retrieving videos in any video retrieval website according to the synonym set;
acquiring a link address of the retrieval video; the link address is a web page address of the retrieved video.
Determining a video webpage corresponding to the link address according to the link address;
deleting the link address when the link address does not have the corresponding video webpage; when the video link address does not have the corresponding video webpage, the video webpage is invalid, and further invalid retrieval links can be screened out.
Triggering videos in the video webpage, responding to the triggering behavior of the video webpage, and acquiring videos of the video webpage to form the retrieval video set when the triggering behavior of the video webpage has a response;
when a response is made, the video corresponding to the video link address is represented, and then the video corresponding to the video webpage with the response can be obtained.
And when the triggering behavior of the video webpage has no response, judging that the video webpage is invalid. When the link address cannot be opened, no webpage exists, and when the video link is effective, a webpage with a false video may exist, and only a webpage corresponding to the video exists, but the webpage does not have a function of loading the corresponding video.
The invention has the beneficial effects that: the video can be screened, and the invalid link address can be deleted, so that the video acquisition efficiency is improved. And through corresponding video response, when the video is valid in response, and when the video is not valid in response, the video is invalid, so that the video acquisition efficiency is further improved.
Example 6:
as an embodiment of the present invention: the retrieving and acquiring the retrieved video set corresponding to the synonym set according to the constraint model further comprises:
acquiring a network transmission protocol of the retrieval video;
judging the acquisition mode of the retrieval video according to the network transmission protocol;
acquiring a corresponding video loading tool according to the acquisition mode of the retrieval video;
and acquiring the retrieval video set according to the loading tool.
The principle of the invention is as follows: the loading of the video is divided into three steps: loading a player, loading an XML file and finally loading a video file; the formats of the videos which can be played by the player are different from the formats of the videos which can be played by the player, such as MPEG, AVI, nAII, ASF, MOV, 3GP, MP4, RMVB and the like, and if the video is played without the player with the relevant format, the video cannot be played. The video downloading mode also needs to be determined when the video is downloaded, for example, the video file is downloaded according to the file position, which needs the P2P technology to identify the MagNet downloading of the file content, or does not need the P2P technology, so that the user can judge whether the video can be downloaded and used through a loading tool which can be used.
The invention has the beneficial effects that: according to the method and the device, whether the video can be used or not can be judged according to whether the video can be opened or not through judging the loading tool. When the video can not be downloaded for use, the information such as the link address of the relevant video is deleted. The waste of resources is prevented, and the connected and downloaded video can be downloaded but cannot be used.
Example 7:
as an embodiment of the present invention: the retrieving and acquiring the retrieved video set corresponding to the synonym set according to the constraint model further comprises:
acquiring a video webpage corresponding to the retrieval video;
judging the frequency of abnormal pop-up of the video webpage;
judging whether the video webpage is an abnormal webpage or not according to the times of the abnormal popup;
when the video webpage is an abnormal webpage, deleting the video webpage;
and when the video webpage is a normal webpage, acquiring the video of the video webpage to form a retrieval video set.
The principle of the invention is as follows: the method comprises the steps of judging whether the webpage has threats or not according to the times of popup windows of the webpage within the preset unit time, and deleting the related webpage when the webpage has abnormal threats.
The invention has the beneficial effects that: according to the number of times of occurrence of the abnormal pop-up windows, whether the threat of Trojan viruses exists in the current video webpage or not can be judged, so that when the video is downloaded, the equipment for downloading the related video is influenced by the viruses, the equipment is damaged, and meanwhile, the acquisition of the video set is also influenced by the damage of the equipment.
Example 8:
as an embodiment of the present invention: the screening of the videos in the retrieval video set according to the characteristics of the synonym set in the video set corresponding to the content through preset characteristic parameters to obtain a target video set includes:
after the retrieval video set is determined, acquiring the semanteme of each synonym in the synonym set;
judging the characteristics of the video corresponding to the synonyms according to the semantic meaning of each synonym in the synonym set;
and determining a final target video set according to the comparison between the characteristics of the videos corresponding to the synonyms and the parameters of the template video.
The principle of the invention is as follows: after the video set is obtained through the synonym, the retrieved video is inaccurate because the semantic meaning of the retrieval word may only appear in one part of the retrieved video set or only the related introduction of the video exists in the semantic meaning of the retrieval word in the video. The video is judged again through the characteristic that the semanteme appears in the video, and the retrieved video is the target video required by the retrieval user only when the characteristic is higher than the threshold value of the characteristic that the preset semanteme appears.
The invention has the beneficial effects that: the method judges whether the searched video is the target video or not according to the characteristics of the semantic appearance of the final search words. Through the final retrieval step, the retrieved video can be made more accurate.
Example 9:
as an embodiment of the present invention: the preset characteristic parameters are determined by the following steps:
step S1: acquiring a retrieval video set E ═ E1,e2L L eg}; the g represents the g video; 1,2,3L g;
step S2: determining a feature matrix R of the retrieval video set according to the retrieval video set E:
wherein, r isglThe ith feature representing the ith video;
step S3: calculating a difference coefficient eta of the retrieval video set according to the characteristic matrix R of the retrieval video set:
wherein the em is a characteristic parameter of the template video, and the N is the number of retrieval video sets;
step S4: determining a threshold value epsilon of the target video as em eta according to the difference coefficient eta and the template video:
when the characteristic parameter of the video in the retrieval video set is larger than or equal to the threshold value of the target video, determining that the video in the retrieval video set is the target video;
when retrieving video e in the video setgWhen the characteristic parameter of the search video set is smaller than the threshold value of the target video, deleting the video e in the search video setg。
The invention has the beneficial effects that: the difference coefficient of each video in the retrieval video set is calculated through the characteristics of the videos in the retrieval video set and the characteristic parameters of the template video. And finally, determining the threshold value of the final target video through the product of the difference coefficient and the template video. Through the threshold value, the finally screened target videos are all matched with the template videos, and then a final target video set is obtained.
Example 10:
as an embodiment of the present invention: in step S2, determining a feature matrix of the search video set includes the following steps:
step S10, obtaining synonym semantics according to synonyms corresponding to the videos in the retrieved video set;
step S11: according to the synonym semantics, performing semantic marking in the video of the retrieval video set;
step S12, extracting the frame image of the semantic mark according to the semantic mark and acquiring the characteristics of the frame image;
step S13: and constructing a feature matrix of the retrieval video set according to the features of the frame images.
The principle of the invention is as follows: according to the invention, the semantics of the synonym are analyzed, the semantic position of the synonym is displayed in the corresponding video in the retrieval video set for semantic marking, then the frame image at the marking position is extracted, and the characteristics are obtained according to the frame image. And finally, according to the feature matrix of the frame image.
The invention has the beneficial effects that: according to the invention, the occurrence frequency of synonym meanings in the retrieval video set can be judged through the semantic mark of the synonym, and the preliminary judgment on the correlation between the retrieval video set and the retrieval word can be preliminarily carried out. And then extracting the frame image, and acquiring the features according to the frame image, without belonging to related features and having no unrelated features to other extracted features in the retrieval video set. The data processing amount is reduced, the efficiency of feature acquisition is improved, and irrelevant features are not required to be judged or processed.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (10)
1. A method for rapidly retrieving a captured video, comprising:
acquiring a search word of a user, and determining a synonym set of the search word;
constructing a constraint model of the retrieval video according to the synonym set, and retrieving and acquiring the retrieval video set corresponding to the synonym set through the constraint model;
and acquiring the characteristics of the corresponding content of the synonym set in the retrieval video set, and screening the videos in the retrieval video set through preset characteristic parameters to obtain a target video set.
2. The method for rapidly searching the collected video according to claim 1, wherein the obtaining the search term of the user and determining the synonym set of the search term comprises:
acquiring a search term;
according to wordsObtaining a first synonym set A ═ { a ═ of the search word1,a2L L ax}；
Acquiring a second synonym set B ═ { B ═ B of the search term according to the webpage label1,b2L L by}；
Acquiring a third synonym set C ═ { C ═ C of the search term according to the text relation1,c2L L cz}；
Performing the same item analysis on the first synonym set A, the second synonym set B and the third synonym set C to obtain a same item set S(A,B,C):
Wherein, the axRepresenting the xth item in the first synonym set A; b isyRepresents the y item in the second synonym set B; c is mentionedzRepresenting the z item in the first synonym set C; 1,2,3L x; 1,2,3L y; 1,2,3, L z;
according to the same item set S(A,B,C)And determining the synonym set.
3. The method for rapidly searching the collected video according to claim 1, wherein the constructing a constraint model of the video according to the synonym set comprises:
step 1: determining the distance D between the synonym set and the search word:
D＝(G-si)P-1(G-si)；
wherein G represents the search term; the P is-1A covariance matrix of the synonym set, siAn ith synonym representing the set of synonyms;
step 2: constructing a linear regression equation of the synonym set;
wherein xi is the inner product of the synonym set; the above-mentioned
Represents a mean of the set of synonyms;
and step 3: determining a constraint model F of the video according to the linear regression equation of the distance D:
wherein, k isjA jth video representing a set of retrieved videos; the above-mentioned
Means representing a set of retrieved videos; the beta represents an error coefficient of the retrieved video set.
4. The method for rapidly retrieving the captured video according to claim 1, wherein the retrieving and acquiring the set of retrieved videos corresponding to the set of synonyms according to the constraint model comprises:
acquiring a first retrieval video set from any video retrieval website according to the synonym set;
determining a mean value of the first search video set;
importing the parameters of the retrieved videos in the retrieved video set into the constraint model according to the average value to obtain a constraint value;
and acquiring a positive value in the constraint value and a video corresponding to the positive value, and determining the retrieval video collection.
5. The method as claimed in claim 1, wherein said retrieving and obtaining the set of retrieved videos corresponding to the set of synonyms according to the constraint model further comprises:
retrieving videos on any video retrieval website according to the synonym set;
acquiring a link address of the retrieval video;
determining a video webpage corresponding to the link address according to the link address;
deleting the link address when the link address does not have the corresponding video webpage;
triggering videos in the video webpage, responding to the triggering behavior of the video webpage, and acquiring videos of the video webpage to form the retrieval video set when the triggering behavior of the video webpage has a response;
and when the triggering behavior of the video webpage has no response, judging that the video webpage is an invalid webpage.
6. The method as claimed in claim 5, wherein said retrieving and obtaining the set of retrieved videos corresponding to the set of synonyms according to the constraint model further comprises:
acquiring a network transmission protocol of the retrieval video;
judging the acquisition mode of the retrieval video according to the network transmission protocol;
acquiring a corresponding video loading tool according to the acquisition mode of the retrieval video;
and acquiring the retrieval video set according to the loading tool.
7. The method as claimed in claim 6, wherein said retrieving and obtaining the set of retrieved videos corresponding to the set of synonyms according to the constraint model further comprises:
acquiring a video webpage corresponding to the retrieval video;
judging the frequency of abnormal pop-up of the video webpage;
judging whether the video webpage is an abnormal webpage or not according to the times of the abnormal popup;
when the video webpage is an abnormal webpage, deleting the video webpage;
and when the video webpage is a normal webpage, acquiring the video of the video webpage to form a retrieval video set.
8. The method as claimed in claim 1, wherein the step of screening videos in the search video set according to the feature of the corresponding content in the search video set by the synonym set by using a preset feature parameter to obtain a target video set comprises:
after the retrieval video set is determined, acquiring the semanteme of each synonym in the synonym set;
judging the characteristics of the video corresponding to the synonyms according to the semantic meaning of each synonym in the synonym set;
and determining a final target video set according to the comparison between the characteristics of the videos corresponding to the synonyms and preset characteristic parameters.
9. The method for rapidly retrieving the captured video according to claim 1, wherein the predetermined characteristic parameter is determined by the following steps:
step S1: acquiring a retrieval video set E ═ E1,e2L L eg}; the g represents the g video; 1,2,3L g;
step S2: determining a feature matrix R of the retrieval video set according to the retrieval video set E:
wherein, r isglThe ith feature representing the ith video;
step S3: calculating a difference coefficient eta of the retrieval video set according to the characteristic matrix R of the retrieval video set:
wherein the em is a characteristic parameter of the template video, and the N is the number of retrieval video sets;
step S4: determining a threshold value epsilon of the target video as em eta according to the difference coefficient eta and the template video:
when the characteristic parameters of the videos in the retrieval video set are larger than or equal to the threshold value of the target video, determining that the videos in the retrieval video set are the target videos;
when retrieving video e in the video setgWhen the characteristic parameter of the search video set is smaller than the threshold value of the target video, deleting the video e in the search video setg。
10. The method for fast searching for captured video according to claim 9, wherein in the step S2, determining the feature matrix of the search video set comprises the following steps:
step S10, obtaining synonym semantics according to synonyms corresponding to the videos in the retrieved video set;
step S11: according to the synonym semantics, performing semantic marking in the video of the retrieval video set;
step S12, extracting the frame image of the semantic mark according to the semantic mark and acquiring the characteristics of the frame image;
step S13: and constructing a feature matrix of the retrieval video set according to the features of the frame images.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011094314.4A CN112269900A (en) | 2020-10-14 | 2020-10-14 | Method for rapidly retrieving collected video |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011094314.4A CN112269900A (en) | 2020-10-14 | 2020-10-14 | Method for rapidly retrieving collected video |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112269900A true CN112269900A (en) | 2021-01-26 |
Family
ID=74337298
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011094314.4A Withdrawn CN112269900A (en) | 2020-10-14 | 2020-10-14 | Method for rapidly retrieving collected video |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112269900A (en) |
-
2020
- 2020-10-14 CN CN202011094314.4A patent/CN112269900A/en not_active Withdrawn
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8630972B2 (en) | Providing context for web articles | |
| US8051080B2 (en) | Contextual ranking of keywords using click data | |
| US8856129B2 (en) | Flexible and scalable structured web data extraction | |
| US8868569B2 (en) | Methods for detecting and removing duplicates in video search results | |
| US20090089278A1 (en) | Techniques for keyword extraction from urls using statistical analysis | |
| JP4797069B2 (en) | Keyword management program, keyword management system, and keyword management method | |
| WO2011080899A1 (en) | Information recommendation method | |
| US8886660B2 (en) | Method and apparatus for tracking a change in a collection of web documents | |
| WO2009129425A2 (en) | Forum web page clustering based on repetitive regions | |
| US7962523B2 (en) | System and method for detecting templates of a website using hyperlink analysis | |
| CN107463616B (en) | Enterprise information analysis method and system | |
| CN110287409B (en) | Webpage type identification method and device | |
| CN110737821B (en) | Similar event query method, device, storage medium and terminal equipment | |
| TWI457775B (en) | Method for sorting and managing websites and electronic device of executing the same | |
| CN112818200A (en) | Data crawling and event analyzing method and system based on static website | |
| CN116015842A (en) | Network attack detection method based on user access behaviors | |
| CN112269906B (en) | Automatic extraction method and device of webpage text | |
| JP2007164633A (en) | Content retrieval method, system thereof, and program thereof | |
| CN111538903B (en) | Method and device for determining search recommended word, electronic equipment and computer readable medium | |
| CN111597386A (en) | Video acquisition method | |
| CN112182451A (en) | Webpage content abstract generation method, equipment, storage medium and device | |
| CN108280102A (en) | Internet behavior recording method, device and user terminal | |
| US11669556B1 (en) | Method and system for document retrieval and exploration augmented by knowledge graphs | |
| CN112269900A (en) | Method for rapidly retrieving collected video | |
| CN114706948A (en) | News processing method and device, storage medium and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20210126 |