阅读说明：本技术 一种源多媒体确定方法、装置、设备及存储介质 (One provenance multimedia determines method, apparatus, equipment and storage medium ) 是由 张晓寒 任可欣 冯知凡 张扬 朱勇 于 2019-09-03 设计创作，主要内容包括：本申请公开了一种源多媒体确定方法、装置、设备及存储介质,涉及智能搜索技术领域。具体实现方案为：获取多媒体片段对应的至少一个多媒体关联信息；其中多媒体关联信息包括多媒体片段的描述文本、媒体元素识别结果、以及用户输入标签信息中的至少一种；对各多媒体关联信息进行实体提取,根据所提取的各实体和预设知识图谱,确定与多媒体片段对应的至少一个关联多媒体；从各关联多媒体中确定与多媒体片段对应的源多媒体。本申请的技术方案通过多媒体关联信息对应的实体和预设知识图谱相结合,确定了关联多媒体,缩小了源多媒体的确定范围,从关联多媒体中进行源多媒体的确定,减少了非真实源多媒体的干扰,提高了源多媒体确定效率和准确率。(This application discloses a provenance multimedias to determine method, apparatus, equipment and storage medium, is related to intelligent search technique field.Specific implementation are as follows: obtain at least one corresponding multimedia related information of multi-media segment；Wherein multimedia related information includes at least one of description text, media elements recognition result and user's input label information of multi-media segment；Entity extraction is carried out to each multimedia related information, according to extracted each entity and default knowledge mapping, determines at least one association multimedia corresponding with multi-media segment；Source corresponding with multi-media segment multimedia is determined from each association multimedia.The technical solution of the application is combined by the corresponding entity of multimedia related information and default knowledge mapping, association multimedia has been determined, reduce the multimedia determining range in source, the multimedia determination in source is carried out from association multimedia, reduce the multimedia interference of non-real actual source, the source multimedia of improving determines efficiency and accuracy rate.)

1. a provenance multimedia determines method characterized by comprising

Obtain at least one corresponding multimedia related information of multi-media segment；Wherein, the multimedia related information includes institute State at least one of the description text, media elements recognition result and user's input label information of multi-media segment；

Entity extraction is carried out to each multimedia related information, according to extracted each entity and default knowledge mapping, is determined At least one association multimedia corresponding with the multi-media segment；

From each association multimedia, source corresponding with multi-media segment multimedia is determined.

2. the method according to claim 1, wherein according to extracted each entity and default knowledge mapping, really Fixed at least one association multimedia corresponding with the multi-media segment, comprising:

According to extracted each entity and the corresponding triplet information of the knowledge mapping, at least one media entities is determined, And using the corresponding multi-medium data of each media entities as association multimedia.

3. according to the method described in claim 2, it is characterized in that, after determining at least one media entities, will be each Before the corresponding multi-medium data of media entities is as association multimedia, the method also includes:

According to the class of the corresponding multimedia related information of the determination frequency of each media entities and/or each media entities Not, the confidence level of each media entities is determined；

Confidence level is filtered out from each media entities meets the media entities to impose a condition；

Correspondingly, using the corresponding multi-medium data of each media entities as association multimedia, comprising:

Using the corresponding multi-medium data of each media entities filtered out as association multimedia.

4. according to the method described in claim 3, it is characterized in that, according to the determination frequency and each multimedia of each media entities The classification of the corresponding multimedia related information of entity, determines the confidence level of each media entities, comprising:

For each media entities, according to the corresponding confidence weight of different classes of multimedia related information, to the multimedia The determination frequency of entity is weighted；

According to the frequency of the media entities after weighting, the corresponding confidence level of the media entities is determined.

5. method according to claim 1-4, which is characterized in that from each association multimedia, determine with The corresponding source multimedia of the multi-media segment, comprising:

The similarity of the finger print information of each multimedia finger print information of association and the multi-media segment is determined respectively；

According to each similarity, source corresponding with multi-media segment multimedia is determined from the association multimedia.

6. according to the method described in claim 5, it is characterized in that, determining each multimedia finger print information of association respectively Before the similarity of the finger print information of the multi-media segment, the method also includes:

Using the corresponding broadcasting link of media entities in the knowledge mapping, the corresponding more matchmakers of the media entities are grabbed Volume data, and extract the finger print information of the multi-medium data；

By the finger print information of the multi-medium data and the media entities associated storage, multimedia fingerprint library is formed.

7. according to the method described in claim 6, it is characterized in that, determine respectively the multimedia finger print information of each association with The similarity of the finger print information of the multi-media segment, comprising:

According to each multimedia media entities of association, searched in the multimedia fingerprint library and each more matchmakers of association The corresponding finger print information of body；

The finger print information of the multi-media segment is extracted, and determines the multimedia finger print information of each association and described more respectively The similarity of the finger print information of media fragment.

8. a provenance multimedia determining device characterized by comprising

Related information obtains module, for obtaining at least one corresponding multimedia related information of multi-media segment；Wherein, described Multimedia related information includes the description text, media elements recognition result and user's input label of the multi-media segment At least one of information；

It is associated with multimedia determining module, for carrying out entity extraction to each multimedia related information, according to extracted each Entity and default knowledge mapping determine at least one association multimedia corresponding with the multi-media segment；

Source multimedia determining module, for from each association multimedia, determining that source corresponding with the multi-media segment is more Media.

9. a kind of electronic equipment characterized by comprising

At least one processor；And

The memory being connect at least one described processor communication；Wherein,

The memory is stored with the instruction that can be executed by least one described processor, and described instruction is by described at least one It manages device to execute, so that at least one described processor is able to carry out provenance multimedia of any of claims 1-7 Determine method.

10. a kind of non-transitory computer-readable storage medium for being stored with computer instruction, which is characterized in that the computer refers to It enables for making the computer perform claim that a provenance multimedia described in any one of 1-7 be required to determine method.

Technical field

This application involves technical field of data processing more particularly to intelligent search technique fields.

Background technique

As multimedia resource (such as audio, video data) increases on internet, there are more and more multimedias relevant Using, such as multi-medium data recommendation, multimedia data searching etc..It is this kind of apply for precisely recommend, search recommend and Precisely when the scenes such as search, need to position multimedia derived data.

Summary of the invention

The embodiment of the present application provides a provenance multimedia and determines method, apparatus, equipment and storage medium, more with the source of improving The determination efficiency of media and the accuracy of definitive result.

In a first aspect, the embodiment of the present application, which provides a provenance multimedia, determines method, comprising:

Obtain at least one corresponding multimedia related information of multi-media segment；Wherein, the multimedia related information packet Include at least one of the description text, media elements recognition result and user's input label information of the multi-media segment；

Entity extraction is carried out to each multimedia related information, according to extracted each entity and default knowledge mapping, Determine at least one association multimedia corresponding with the multi-media segment；

From each association multimedia, source corresponding with multi-media segment multimedia is determined.

One embodiment in above-mentioned application passes through the acquisition of the corresponding multimedia related information of multi-media segment, and extracts The entity of multimedia related information is associated multimedia screening, Jin Ercong according to extracted entity and default knowledge mapping It is associated in multimedia and determines source corresponding with multi-media segment multimedia, it is long to solve source multimedia determination process short time consumption, Source multimedia determines the lower problem of accuracy.The technical solution of the application passes through the multimedia related information pair under different dimensions The entity and default knowledge mapping answered combine, it is determined that association multimedia reduces the multimedia determining range in source, from association The multimedia determination in source is carried out in multimedia, reduces the multimedia interference of non-real actual source, improves the multimedia determining effect in source Rate and accuracy rate.

Optionally, according to extracted each entity and default knowledge mapping, determination is corresponding with the multi-media segment extremely A few association multimedia, comprising:

According to extracted each entity and the corresponding triplet information of the knowledge mapping, at least one multimedia reality is determined Body, and using the corresponding multi-medium data of each media entities as association multimedia.

One embodiment in above-mentioned application by extracted each entity and the corresponding triplet information of knowledge mapping into The determination of row media entities expands extracted entity by knowledge mapping, further excavation and multi-media segment Corresponding entity information；Non-multimedia entity in entity after expansion is rejected, identified pass is further reduced Join multimedia quantity, improves the degree of correlation between association multimedia source multimedia corresponding with multi-media segment.

Optionally, after determining at least one media entities, by the corresponding multi-medium data of each media entities Before association multimedia, the method also includes:

According to the corresponding multimedia related information of the determination frequency of each media entities and/or each media entities Classification determines the confidence level of each media entities；

Confidence level is filtered out from each media entities meets the media entities to impose a condition；

Correspondingly, using the corresponding multi-medium data of each media entities as association multimedia, comprising:

Using the corresponding multi-medium data of each media entities filtered out as association multimedia.

One embodiment in above-mentioned application is corresponding by the determination frequency of each media entities and/or each media entities File association information classification, carry out the determination of the confidence level of media entities, and by identified confidence level to more matchmakers Body entity is screened, and the quantity of media entities is reduced, and has further reduced the identified multimedia quantity of association, Connect the degree of correlation improved between association multimedia source multimedia corresponding with multi-media segment.

Optionally, letter is associated with according to the determination frequency of each media entities multimedia corresponding with each media entities The classification of breath determines the confidence level of each media entities, comprising:

For each media entities, according to the corresponding confidence weight of different classes of multimedia related information, to described more The determination frequency of media entities is weighted；

According to the frequency of the media entities after weighting, the corresponding confidence level of the media entities is determined.

One embodiment in above-mentioned application is real according to the determination frequency of each media entities and each multimedia by refinement The classification of the corresponding multimedia related information of body, carries out the determination of confidence level, the perfect determination mechanism of confidence level, from determining frequency Secondary and multimedia related information two dimensions of classification carry out the determination of confidence level, improve between confidence level and multi-media segment Relevance so that the confidence level of media entities is higher, the corresponding multi-medium data of the media entities is multi-media segment The corresponding multimedia possibility in source is bigger, improves indirectly related between identified association multimedia and multi-media segment Degree.

Optionally, from each association multimedia, source corresponding with multi-media segment multimedia is determined, comprising:

The similarity of the finger print information of each multimedia finger print information of association and the multi-media segment is determined respectively；

According to each similarity, the more matchmakers in source corresponding with the multi-media segment are determined from the association multimedia Body.

One embodiment in above-mentioned application passes through the determination of finger print information similarity, is associated multimedia and multimedia The comparison of segment, and then corresponding with the multi-media segment multimedia determination in source is carried out from association multimedia, perfect source are more The determination mechanism of media, while further improving the accuracy of the multimedia definitive result in source.

Optionally, in the finger print information for determining each the association multimedia finger print information and the multi-media segment respectively Similarity before, the method also includes:

Using the corresponding broadcasting link of media entities in the knowledge mapping, it is corresponding to grab the media entities Multi-medium data, and extract the finger print information of the multi-medium data；

By the finger print information of the multi-medium data and the media entities associated storage, multimedia fingerprint library is formed.

Optionally, the finger print information of each multimedia finger print information of association and the multi-media segment is determined respectively Similarity, comprising:

According to each multimedia media entities of association, searched and each association in the multimedia fingerprint library The corresponding finger print information of multimedia；

The finger print information of the multi-media segment is extracted, and determines the multimedia finger print information of each association and institute respectively State the similarity of the finger print information of multi-media segment.

One embodiment in above-mentioned application avoids the more matchmakers in carry out source by the building in progress multimedia fingerprint library in advance The extraction operation that finger print information is carried out in body determination process, to reduce data operation quantity when source multimedia determines, into one Step improves the multimedia determining efficiency in source.

Second aspect, the embodiment of the present application also provides a provenance multimedia determining devices, comprising:

Related information obtains module, for obtaining at least one corresponding multimedia related information of multi-media segment；Wherein, The multimedia related information includes description text, media elements recognition result and the user input of the multi-media segment At least one of label information；

It is associated with multimedia determining module, for carrying out entity extraction to each multimedia related information, according to being extracted Each entity and default knowledge mapping, determine it is corresponding with the multi-media segment at least one be associated with multimedia；

Source multimedia determining module, for from each association multimedia, determination to be corresponding with the multi-media segment Source multimedia.

The third aspect, the embodiment of the present application also provides a kind of electronic equipment, comprising:

At least one processor；And

The memory being connect at least one described processor communication；Wherein,

The memory is stored with the instruction that can be executed by least one described processor, and described instruction is by described at least one A processor executes, so that at least one described processor is able to carry out a provenance multimedia provided by first aspect embodiment Determine method.

Fourth aspect is stored with the non-instantaneous of computer instruction and computer-readable deposits the embodiment of the present application also provides a kind of Storage media, the computer instruction are true for making the computer execute a provenance multimedia provided by first aspect embodiment Determine method.

Other effects possessed by above-mentioned optional way are illustrated hereinafter in conjunction with specific embodiment.

Detailed description of the invention

Attached drawing does not constitute the restriction to the application for more fully understanding this programme.Wherein:

Fig. 1 is the flow chart that one of the embodiment of the present application one source multimedia determines method；

Fig. 2 is the flow chart that one of the embodiment of the present application two source multimedia determines method；

Fig. 3 is the structure chart of one of the embodiment of the present application three source multimedia determining device；

Fig. 4 is for realizing that the source multimedia of the embodiment of the present application determines the block diagram of the electronic equipment of method.

Specific embodiment

It explains below in conjunction with exemplary embodiment of the attached drawing to the application, including the various of the embodiment of the present application Details should think them only exemplary to help understanding.Therefore, those of ordinary skill in the art should recognize It arrives, it can be with various changes and modifications are made to the embodiments described herein, without departing from the scope and spirit of the present application.Together Sample, for clarity and conciseness, descriptions of well-known functions and structures are omitted from the following description.