US20140207778A1

US20140207778A1 - System and methods thereof for generation of taxonomies based on an analysis of multimedia content elements

Info

Publication number: US20140207778A1
Application number: US14/224,923
Authority: US
Inventors: Igal RAICHELGAUZ; Karina ODINAEV; Yehoshua Y. Zeevi
Original assignee: Cortica Ltd
Current assignee: Cortica Ltd
Priority date: 2005-10-26
Filing date: 2014-03-25
Publication date: 2014-07-24
Also published as: US20200175550A1; US9798795B2; US20130138653A1; US20150332154A1; US20170270107A1; US20090112864A1; US20170140029A1; US20150019586A1; US10552380B2; US11657079B2; US20160321256A1; US11238066B2; US20090282218A1; US20200167314A1; US9009086B2; US20200193308A1; US20200183965A1; US8386400B2; US20200125837A1; US20150234851A1

Abstract

A method and system for generating concept structures and taxonomies based on received multimedia data elements (MMDEs) are provided. The method comprises receiving at least one MMDE; generating at least one signature for the at least one received MMDE; matching the at least one generated signature to a plurality of clusters to find at least one matching cluster; associating the at least one generated signature with each of the at least one matching cluster; and analyzing the at least one generated signature with respect to a signature reduced cluster (SRC) of each of the at least one matching cluster to generate a taxonomy, wherein the taxonomy relates to the at least one received MMDE and an MMDE respective of each of the at least one matching cluster.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

This application claims the benefit of U.S. Provisional Application No. 61/805,158 filed on Mar. 26, 2013. This application is also a continuation-in-part of:
(1) U.S. patent application Ser. No. 14/175,569, filed Feb. 7, 2014, now pending, which is a continuation of U.S. patent application Ser. No. 12/084,150 having a filing date of Apr. 7, 2009, now U.S. Pat. No. 8,655,801, which is the National Stage of International Application No. PCT/IL2006/001235, filed on Oct. 26, 2006, which claims foreign priority from Israeli Application No. 171577 filed on Oct. 26, 2005 and Israeli Application No. 173409 filed on 29 Jan. 2006;
(2) U.S. patent application Ser. No. 13/682,132, filed Nov. 20, 2012, which is a continuation of U.S. patent application Ser. No. 12/195,863, filed Aug. 21, 2008, now U.S. Pat. No. 8,326,775, which claims priority under 35 USC 119 from Israeli Application No. 185414, filed on Aug. 21, 2007, and which is also a continuation-in-part of the above-referenced U.S. patent application Ser. No. 12/084,150;
(3) U.S. patent application Ser. No. 12/348,888, filed Jan. 5, 2009, now pending, which is a CIP of U.S. patent application Ser. No. 12/084,150, filed on Apr. 7, 2009 and U.S. patent application Ser. No. 12/195,863 filed on Aug. 21, 2008; and
(4) U.S. patent application Ser. No. 13/668,559, filed Nov. 5, 2012, which is a continuation of U.S. patent application Ser. No. 12/538,495, filed Aug. 10, 2009, now U.S. Pat. No. 8,312,031, which is a CIP of U.S. patent application Ser. No. 12/084,150 filed on Apr. 7, 2009, U.S. patent application Ser. No. 12/195,863, filed on Aug. 21, 2008; and U.S. patent application Ser. No. 12/348,888, filed Jan. 5, 2009. All of the applications referenced above are herein incorporated by reference.

TECHNICAL FIELD

The invention relates generally to content-management and search engines and, more particularly, to the collection, clustering, and creation of concept structures of multimedia data elements for the purpose of effective storage, management, knowledge database generation, and search.

BACKGROUND

With the abundance of multimedia data made available through various means in general and, in particular, through the Internet and world-wide web (WWW), there is a need for effective ways of searching for and managing such multimedia data. Searching, organizing and managing multimedia data in general and, in particular, video data may be challenging at best due to the difficulties associated with representing and comparing the information embedded in the video content, and further due to the scale of information that needs to be checked. Moreover, when it is necessary to find a content of a video by means of textual query, prior art solutions revert to analyzing various metadata that textually describe the content of the multimedia data. However, such content may be abstract and complex by nature, and is not necessarily adequately defined by the existing and/or attached metadata.
The rapid increase in multimedia databases, accessible for example through the Internet, calls for the application of new methods of representing information embedded in video content. Searching for multimedia in general and for video data in particular is challenging due to the huge amount of information that must have been previously indexed, classified, and clustered. Moreover, prior art techniques revert to model-based methods to define and/or describe multimedia data. However, by its very nature, the structure of such multimedia data may be too abstract and/or complex to be adequately represented by metadata. The greatest difficulty arises in cases where the target sought for multimedia data is not adequately defined in words or by respective metadata of the multimedia data. For example, it may be desirable to locate a car of a particular model in a large database of video clips or segments. In some cases, the model of the car would be part of the metadata, but in many cases it would not. Moreover, the car may be shown at angles different from the angles of a specific photograph of the car that is available as a search item. Similarly, if a piece of music, as in a sequence of notes, is to be found, it is not necessarily the case that in all available content the notes are known in their metadata form, or for that matter, the search pattern may just be a brief audio clip.
A system implementing a computational architecture (hereinafter “the Architecture”) that is based on a PCT patent application publication number WO2007/049282 and published on May 3, 2007, entitled “A Computing Device, a System and a Method for Parallel Processing of Data Streams”, assigned to common assignee, is hereby incorporated by reference for all the useful information it contains. Generally, the Architecture consists of a large ensemble of randomly, independently generated, heterogeneous processing cores, mapping in parallel data-segments onto a high-dimensional space and generating compact signatures for classes of interest.
Searching multimedia data has been a challenge in the past several years and has therefore received considerable attention. Early systems would take a multimedia data element in the form of, for example, an image, compute various visual features from it and then search one or more indexes to return images with similar features. In addition, values for these features and appropriate weights reflecting their relative importance could be also used. Searching and indexing techniques have improved over time to improve handling of various types of multimedia inputs. However, due to the exponential growth of the use of the Internet and the multimedia data available there, these prior art systems have become less effective in handling the multimedia data, due to the vast amounts already existing, as well as the speed at which new ones are added.
Searching has therefore become a significant challenge and even the addition of metadata to assist in the search has limited functionality. Firstly, metadata may be inaccurate or not fully descriptive of the content contained within the multimedia data, and secondly, not every piece of multimedia data can be described accurately enough by a sequence of textual metadata. A query model for a search engine has some advantages, such as comparison and ranking of images based on objective visual features, rather than on subjective image annotations. However, the query model has its drawbacks as well. Certainly when no metadata is available and only the multimedia data needs to be used, the process requires significant effort. Those skilled in the art will appreciate that there is no known intuitive and automated way of describing multimedia data in a manner consistent with an average user's conception of the data. Therefore, a user's perception or conceptual understanding of the multimedia data may significantly differ from the way it is actually stored and manipulated by a search engine.
The current generation of web applications has become more and more effective at aggregating massive amounts of data of different multimedia content, such as, pictures, videos, clips, paintings and mash-ups. Such applications are capable of slicing and dicing components of multimedia content in different ways, as well as searching the content and displaying it in an organized fashion by using, for example, concept networks. A concept network may enable understanding of a multimedia data based on its related concept. However, current art is unable to add any real “intelligence” to the mix, i.e., no new knowledge is extracted from the multimedia data that are aggregated by such systems. Moreover, the systems tend to be non-scalable due to the vast amounts of data they have to handle. This inherently hinders these systems' ability to provide high quality searching for multimedia content.

SUMMARY

Certain embodiments disclosed herein include a method for generating taxonomies based on analysis of at least one multimedia data element (MMDE). The method comprises receiving the at least on MMDE; generating at least one signature for the at least one received MMDE; matching the at least one generated signature to a plurality of clusters to find at least one matching cluster; associating the at least one generated signature with each of the at least one matching cluster; and analyzing the at least one generated signature with respect to a signature reduced cluster (SRC) of each of the at least one matching cluster to generate a taxonomy, wherein the taxonomy relates to the at least one received MMDE and an MMDE respective of each of the at least one matching cluster.
Certain embodiments disclosed herein also include a system for generating concept structures. The system comprises an interface to a network for receiving a request to create a new concept structure, wherein the request includes at least a multimedia data element (MMDE) related to the new concept structure; a processor; and a memory connected to the processor, the memory contains instructions that when executed by the processor, configure the system to: receive the at least one MMDE; generate at least one signature for the at least one received MMDE; match the at least one generated signature to a plurality of clusters to find at least one matching cluster; associate the at least one generated signature with each of the at least one matching cluster; and analyze the at least one generated signature with respect to a signature reduced cluster (SRC) of each of the at least one matching cluster to generate a taxonomy, wherein the taxonomy relates to the at least one received MMDE and an MMDE respective of each of the at least one matching cluster.

BRIEF DESCRIPTION OF THE DRAWINGS

The subject matter disclosed herein is particularly pointed out and distinctly claimed in the claims at the conclusion of the specification. The foregoing and other objects, features and advantages of the disclosed embodiments will be apparent from the following detailed description taken in conjunction with the accompanying drawings.

FIG. 1 is a diagram of a DCC system for creating concept structures.

FIG. 2 is a flowchart illustrating the operation of the patch attention processor of the DCC system.

FIG. 3 is a block diagram depicting the basic flow of information in a large-scale video matching system.

FIG. 4 is a diagram showing the flow of patches generation, response vector generation, and signature generation in a large-scale speech-to-text system.

FIG. 5 is a flowchart illustrating the operation of the clustering processor of the DCC system.

FIG. 6 is a flowchart illustrating the operation of the concept generator of the DCC system.

FIG. 7 is a flowchart illustrating the operation of taxonomies generation in accordance with an embodiment.

DETAILED DESCRIPTION

The embodiments disclosed herein are only examples of the many possible advantageous uses and implementations of the innovative teachings presented herein. In general, statements made in the specification of the present application do not necessarily limit any of the various claimed inventions. Moreover, some statements may apply to some inventive features but not to others. In general, unless otherwise indicated, singular elements may be in plural and vice versa with no loss of generality. In the drawings, like numerals refer to like parts through several views.
A large-scale web-platform for a multimedia deep-content-classification (DCC) system configured to continuously create a knowledge database for multimedia data can be utilized to achieve the embodiments disclosed herein, and specifically for generating of taxonomies. The DCC system initially receives a large number of multimedia data elements (MMDEs) to create a knowledge base that is condensed into concept structures that are efficient to store, retrieve and check for matches. As new MMDEs are collected they are efficiently added to the knowledge base and concept structures, such that the computing resources requirement for achieving this operation is generally sub-linear rather than linear or exponential. The DCC system extracts patterns from each MMDE and selects the important/salient patterns for the creation of signatures thereof. A process of inter-matching between the patterns found by clustering is followed by reduction in the number of signatures in a cluster to a minimum that maintains matching and enables generalization to new MMDEs. Metadata respective of the MMDEs is thereby produced, forming together with the reduced clusters into a concept structure.
FIG. 1 shows an exemplary and non-limiting diagram of a DCC system 100 for creating concept structures. The DCC system 100 is configured to receive multimedia data elements (MMDEs), for example from the Internet via the network interface 160. The MMDEs include, but are not limited to, images, graphics, video streams, video clips, audio streams, audio clips, video frames, photographs, images of signals, combinations thereof, and portions thereof. The images of signals are images such as, but not limited to, medical signals, geophysical signals, subsonic signals, supersonic signals, electromagnetic signals, and infrared signals.
The MMDEs may be stored in a database (DB) 150 or kept in the DB 150 for future retrieval of the respective multimedia data element. Such a reference may be, but is not limited to, a universal resource locator (URL). Every MMDE in the DB 150, or referenced therefrom, is then processed by a patch attention processor (PAP) 110 resulting in a plurality of patches that are of specific interest, or otherwise of higher interest than other patches. A more general pattern extraction, such as an attention processor (AP) may also be used in lieu of patches. The AP receives the MMDE that is partitioned into items; an item may be an extracted pattern or a patch, or any other applicable partition depending on the type of the MMDE. The functions of the PAP 110 are described herein below in more detail.
Those patches that are of higher interest are then used by a signature generator (SG) 120 to generate signatures respective of the patch. The operation of the signature generator (SG) 120 is described in more detail herein below. A clustering process (CP) 130 initiates a process of inter-matching of the signatures once it determines that there are a number of patches that are above a predefined threshold. The threshold may be defined to be large enough to enable proper and meaningful clustering. With a plurality of clusters a process of clustering reduction takes place so as to extract the most useful data about the cluster and keep it at an optimal size to produce meaningful results. The process of cluster reduction is continuous. When new signatures are provided after the initial phase of the operation of the CP 130, the new signatures may be immediately checked against the reduced clusters to save on the operation of the CP 130. A more detailed description of the operation of the CP 130 is provided herein below.
A concept generator (CG) 140 operates to create concept structures from the reduced clusters provided by the CP 130. Each concept structure comprises a plurality of metadata associated with the reduced clusters. The result is a compact representation of a concept that can now be easily compared against a MMDE to determine if the received MMDE matches a concept structure stored, for example in the DB 150, by the CG 140. This can be done, for example and without limitation, by providing a query to the DCC system 100 for finding a match between a concept structure and a MMDE. A more detailed description of the operation of the CG 140 is provided herein below.
It should be appreciated that the DCC system 100 can generate a number of concept structures significantly smaller than the number of MMDEs. For example, if one billion (109) MMDEs need to be checked for a match against another one billon MMDEs, typically the result is that no less than 10̂9×10̂9=10̂18 matches have to take place, a daunting undertaking. The DCC system 100 would typically have around 10 million concept structures or less, and therefore at most only 2×10̂6×10̂9=2×10̂15 comparisons need to take place, a mere 0.2% of the number of matches that have had to be made by other solutions. As the number of concept structures grows significantly slower than the number of MMDEs, the advantages of the DCC system 100 would be apparent to one with ordinary skill in the art.
The operation of the PAP 110 will now be provided in greater detail with respect to an image as the MMDE. However, this should not be understood as to limit the scope of the invention; other types of MMDEs are specifically included herein and may be handled by the PAP 110.
FIG. 2 depicts an exemplary and non-limiting flowchart 200 of the operation of the PAP 110. In S210, the PAP 110 receives a MMDE from a source for such MMDEs. Such a source may be a system that feeds the DCC system 100 with MMDEs or other sources for MMDEs, for example the world-wide-web (WWW). In S220, the PAP 110 creates a plurality of patches from the MMDE. A patch of an image is defined by, for example, its size, scale, location and orientation. A patch may be, for example and without limitation, a portion of an image of a size 20 pixels by 20 pixels of an image that is 1,000 pixels by 500 pixels. In the case of audio, a patch may be a segment of audio 0.5 seconds in length from a 5 minute audio clip. In S230, a patch not previously checked is processed to determine its entropy. The entropy is a measure of the amount of interesting information that may be present in the patch. For example, a continuous color of the patch has little interest whereas sharp edges, corners or borders, will result in higher entropy representing a lot of interesting information. The plurality of statistically independent cores, the operation of which is discussed in more detail herein below, is used to determine the level-of-interest of the image; a process of voting takes place to determine whether the patch is of interest or not.
In S240, it is checked whether the entropy was determined to be above a predefined threshold, and if so execution continues with S250; otherwise, execution continues with S260. In S250 the patch having entropy above the threshold is stored for future use by the SG 120 in, for example, DB 150. In S260 it is checked whether there are more patches of the MMDE to be checked, and if so execution continues with S220; otherwise execution continues with S270. In S270 it is checked whether there are additional MMDEs, and if so execution continues with S210; otherwise, execution terminates. It would be appreciated by those of skill in the art that this process reduces the information that must be handled by the DCC system 100 by focusing on areas of interest in the MMDEs rather than areas that are less meaningful for the formation of a concept structure.
A high-level description of the process for large scale video matching performed by the Matching System is depicted in FIG. 3. Video content segments 2 from a Master DB 6 and a Target DB 1 are processed in parallel by a large number of independent computational Cores 3 that constitute the Architecture. Further details on the computational Cores generation are provided below. The independent Cores 3 generate a database of Robust Signatures and Signatures 4 for Target content-segments 5 and a database of Robust Signatures and Signatures 7 for Master content-segments 8. An exemplary and non-limiting process of signature generation for an audio component is shown in detail in FIG. 4. Referring back to FIG. 3, at the final step, Target Robust Signatures and/or Signatures are effectively matched, by a matching algorithm 9, to Master Robust Signatures and/or Signatures database to find all matches between the two databases.
A brief description of the operation of the SG 120 is therefore provided, this time with respect to a MMDE which is a sound clip. However, this should not be understood as to limit the scope of the invention and other types of MMDEs are specifically included herein and may be handled by SG 120. To demonstrate an example of signature generation process, it is assumed, merely for the sake of simplicity and without limitation on the generality of the disclosed embodiments, that the signatures are based on a single frame, leading to certain simplification of the computational core's generation. The Matching System shown in FIG. 3 is extensible for signatures generation capturing the dynamics in-between the frames and the information of the frame's patches.
The signatures generation process will be described with reference to FIG. 4. The first step in the process of signatures generation from a given speech-segment is to break-down the speech-segment to K patches 14 of random length P and random position within the speech segment 12. The break-down is performed by the patch generator component 21. The value of K is determined based on optimization, considering the tradeoff between accuracy rate and the number of fast matches required in the flow process of the Matching System. In the next step, all the K patches are injected in parallel to all L computational Cores 3 to generate K response vectors 22. The vectors 22 are fed into the SG 120 to produce a Signatures and Robust Signatures 4.
In order to generate Robust Signatures, i.e., Signatures that are robust to additive noise L (where L is an integer equal to or greater than 1) computational cores are utilized in the Matching System. A frame i is injected into all the Cores. The computational cores 3 generate two binary response vectors: {right arrow over (S)} which is a Signature vector, and {right arrow over (RS)} which is a Robust Signature vector.
For generation of signatures robust to additive noise, such as White-Gaussian-Noise, scratch, etc., but not robust to distortions, such as crop, shift and rotation, etc., a core C_i{n_i} (1≦i≦L) may consist of a single leaky integrate-to-threshold unit (LTU) node or more nodes. The node ni equations are:
$V_{i} \sum_{j} w_{ij} k_{j}$
ni=□(Vi−Thx); □ is a Heaviside step function; w_gis a coupling node unit (CNU) between node i and image component j (for example, grayscale value of a certain pixel j); k_jis an image component j (for example, grayscale value of a certain pixel j); Th_xis a constant Threshold value, where x is ‘S’ for Signature and ‘RS’ for Robust Signature; and V_iis a Coupling Node Value.
The Threshold values Th_xare set differently for Signature generation and for Robust Signature generation. For example, for a certain distribution of V_ivalues (for the set of nodes), the thresholds for Signature (ThS) and Robust Signature (ThRS) are set apart, after optimization, according to at least one or more of the following criteria:
For: V _i>Th_RS
1−p(V>Th_S)−1−(1−ε)^l<<1 I:
i.e., given that I nodes (cores) constitute a Robust Signature of a certain image I, the probability that not all of these I nodes will belong to the Signature of same, but noisy image,
is sufficiently low (according to a system's specified accuracy).
p(V _i>Th_RS)≈l/L II:
i.e., approximately 1 out of the total L nodes can be found to generate Robust Signature according to the above definition.
Both Robust Signature and Signature are generated for certain frame i. III:
It should be understood that the creation of a signature is a unidirectional compression where the characteristics of the compressed data are maintained but the compressed data cannot be reconstructed. Therefore, a signature can be used for the purpose of comparison to another signature without the need of comparison of the original data. The detailed description of the Signature generation can be found U.S. Pat. Nos. 8,326,775 and 8,312,031, assigned to common assignee, which are hereby incorporated by reference for all the useful information they contain.
Computational Core generation is a process of definition, selection and tuning of the Architecture parameters for a certain realization in a specific system and application. The process is based on several design considerations, such as: (a) The Cores should be designed so as to obtain maximal independence, i.e. the projection from a signal space should generate a maximal pair-wise distance between any two Cores' projections into a high-dimensional space; (b) The Cores should be optimally designed for the type of signals, i.e. the Cores should be maximally sensitive to the spatio-temporal structure of the injected signal, for example, and in particular, sensitive to local correlations in time and space. Thus, in some cases a Core represents a dynamic system, such as in state space, phase space, edge of chaos, etc., which is uniquely used herein to exploit their maximal computational power, and, (c) The Cores should be optimally designed with regard to invariance to a set of signal distortions, of interest in relevant applications. Detailed description of the Computational Core generation, the computational architecture, and the process for configuring such cores is discussed in more detail in U.S. Pat. No. 8,655,801 referenced above.
Hence, signatures are generated by the SG 120 responsive of patches received either from the PAP 110, or retrieved from the DB 150, as discussed hereinabove. It should be noted that other ways for generating signatures may also be used for the purpose the DCC system 100. Furthermore, as noted above, the array of computational cores may be used by the PAP 110 for the purpose of determining if a patch has an entropy level that is of interest for signature generation according to the principles of the invention. The generated signatures are stored, for example, in the DB 150, with reference to the MMDE and the patch for which it was generated thereby enabling back annotation as may be necessary.
Portions of the CP 130 have been discussed in detail in the U.S. Pat. No. 8,386,400, entitled “Unsupervised Clustering of Multimedia Data Using a Large-Scale Matching System”, assigned to common assignee (the “'400 Patent”), and which is hereby incorporated for all that it contains. In accordance with an embodiment an inter-match process and clustering thereof is utilized. The process can be performed on signatures provided by the SG 120. It should be noted though that this inter-matching and clustering process is merely an example for the operation of the CP 130 and other inter-matching and/or clustering processes may be used for the purpose of the invention.
Following is a brief description of the inter-match and clustering process. The unsupervised clustering process maps a certain content-universe onto a hierarchical structure of clusters. The content-elements of the content-universe are mapped to signatures, when applicable. The signatures of all the content-elements are matched to each other, and consequently generate the inter-match matrix. The described clustering process leads to a set of clusters. Each cluster is represented by a small/compressed number of signatures, for example signatures generated by SG 120 as further explained hereinabove, which can be increased by variants. This results in a highly compressed representation of the content-universe. A connection graph between the multimedia data elements of a cluster may be stored. The graph can then be used to assist a user searching for data to move along the graph in the search of a desired multimedia data element.
In another embodiment, upon determination of a cluster, a signature for the whole cluster may be generated based on the signatures of the multimedia data elements that belong to the cluster. It should be appreciated that using a Bloom filter may be used to reach such signatures. Furthermore, as the signatures are correlated to some extent, the hash functions of the Bloom filter may be replaced by simpler pattern detectors, with the Bloom filter being the upper limit.
While signatures are used here as the basic data elements, it should be realized that other data elements may be clustered using the techniques discussed above. For example, a system generating data items is used, where the data items generated may be clustered according to the disclosed principles. Such data items may be, without limitation, multimedia data elements. The clustering process may be performed by dedicated hardware or by using a computing device having storage to store the data items generated by the system and then performing the process described herein above. Then, the clusters can be stored in memory for use as may be deemed necessary.
The CP 130 further uses an engine designed to reduce the number of signatures used in a structure, in a sense, extracting only the most meaningful signatures that identify the cluster uniquely. This can be done by testing a removal of a signature from a cluster and checking if the MMDEs associated with the cluster are still capable of being recognized by the cluster through signature matching.
The process of signature extraction is on-going as the DCC system 100 operates. It should be noted that after initialization, upon signature generation by the SG 120 of a MMDE, its respective signature is first checked against the clusters to see if there is a match and if so it may not be necessary to add the signature to the cluster or clusters but rather simply by associating the MMDE with the identified cluster or clusters. However, in some cases where additional refinement of the concept structure is possible, the signature may be added, or at times even replace one or more of the existing signatures in the reduced cluster. If no match is found then the process of inter-matching and clustering may take place.
FIG. 5 depicts an exemplary and non-limiting flowchart 500 of the operation of the CP 130. In S510 a signature of a MMDE is received, for example from the SG 120. In S520 it is checked whether the signature matches one or more existing clusters and if so execution continues with S550; otherwise, execution continues with S530. In S530 an inter-match between a plurality of signatures previously received by the DCC system 100 is performed, for example in accordance with the principles of the '400 patent. As may be necessary the DB 150 may be used to store results or intermediate results as the case may be, however, other memory elements may be used. In S540 a clustering process takes place, for example in accordance with the principles of the '400 Patent. As may be necessary the DB 150 may be used to store results or intermediate results as the case may be, however, other memory elements may be used.
In S550, the signature identified to match one or more clusters is associated with the existing cluster(s). In S560 it is checked whether a periodic cluster reduction is to be performed, and if so execution continues with S570; otherwise, execution continues with S580. In S570 the cluster reduction process is performed. Specifically, the purpose of the operation is to ensure that in the cluster there remains the minimal number of signatures that still identify all of the MMDEs that are associated with the signature reduced cluster (SRC). This can be performed, for example, by attempting to match the signatures of each of the MMDEs associated with the SRC having one or more signatures removed therefrom. The process of cluster reduction for the purpose of generating SRCs may be performed in parallel and independently of the process described herein above. In such a case after either S560 or S570 the operation of S580 takes place. In S580 it is checked whether there are additional signatures to be processed and if so execution continues with S510; otherwise, execution terminates. SRCs may be stored in memory, such as DB 150, for the purpose of being used by other elements comprising the DCC system 100.
The CG 140 performs two tasks, it associates metadata to the SRCs provided by the CP 130 and it associates between similar clusters based on commonality of metadata. Exemplary and non-limiting methods for associating metadata with MMDEs is described in U.S. patent application Ser. No. 12/348,888, entitled “Methods for Identifying Relevant Metadata for Multimedia Data of a Large-Scale Matching System”, filed on Jan. 5, 2009, assigned to common assignee (the “'888 Application”), and which is hereby incorporated for all that it contains. One embodiment of the '888 Application includes a method for identifying and associating metadata to input MMDEs. The method comprises comparing an input first MMDE to at least a second MMDE; collecting metadata of at least the second MMDE when a match is found between the first MMDE and at least the second MMDE; associating at least a subset of the collected metadata to the first MMDE; and storing the first MMDE and the associated metadata in a storage.
Another embodiment of the '888 Application includes a system for collecting metadata for a first MMDE. The system comprises a plurality of computational cores enabled to receive the first MMDE, each core having properties to be statistically independent of each other core, each generate responsive to the first MMDE a first signature element and a second signature element, the first signature element being a robust signature; a storage unit for storing at least a second MMDE, metadata associated with the second MMDE, and at least one of a first signature and a second signature associated with the second MMDE, the first signature being a robust signature; and a comparison unit for comparing signatures of MMDEs coupled to the plurality of computational cores and further coupled to the storage unit for the purpose of determining matches between multimedia data elements; wherein responsive to receiving the first MMDE the plurality of computational cores generate a respective first signature of said first MMDE and/or a second signature of said first MMDE, for the purpose of determining a match with at least a second MMDE stored in the storage and associating metadata associated with the at least second MMDE with the first MMDE.
Similar processes to match metadata with a MMDE or signatures thereof may be used. Accordingly, each SRC is associated with metadata which is the combination of the metadata associated with each of the signatures that are included in the respective SRC, preferably without repetition of metadata. A plurality of SRCs having metadata may now be associated to each other based on the metadata and/or partial match of signatures. For example, and without limitation, if the metadata of a first SRC and the metadata of a second SRC overlap more than a predetermined threshold level, for example 50% of the metadata match, they may be considered associated clusters that form a concept structure. Similarly, a second threshold level can be used to determine if there is an association between two SRCs where at least a number of signatures above the second threshold are identified as a match with another SRC. As a practical example one may want to consider the concept of Abraham Lincoln where images of the late President and features thereof, appear in a large variety of photographs, drawings, paintings, sculptures and more and are associated as a concept structure of the concept “Abraham Lincoln”. Each concept structure may be then stored in memory, for example, the DB 150 for further use.
FIG. 6 shows an exemplary and non-limiting flowchart 600 of the operation of the CG 140. In S610 the CG 140 receives a SRC from either the CP 130 or by accessing memory, for example, the DB 150. In S620 metadata are generated for the signatures of the SRC, for example in accordance with the principles described hereinabove. A list of the metadata is created for the SRC preferably with no metadata duplication. In one embodiment the commonality of metadata is used to signify the strength of the metadata with respect to a signature and/or the SRC, i.e., a higher number of metadata repetitions is of more importance to the SRC than a lower number of repetitions. Furthermore, in one embodiment a threshold may be used to remove those metadata that have a significantly low rate of repetition as not being representative of the SRC.
In S630 the SRC is matched to previously generated SRCs to attempt to find various matches, as described, for example, hereinabove in more detail. In S640, it is checked if at least one match was found and if so, execution continues with S650; otherwise, execution continues with S660. In S650 the SRC is associated with one or more of the concept structures to which the SRC has been shown to match. In S660 it is checked whether additional SRCs are to be received and if so execution continues with S610; otherwise, execution terminates.
A person skilled in the art would now appreciate the advantages of the DCC system 100 and methods thereof. The DCC system 100 is capable of creating automatically and in an unsupervised fashion concept structures of a wide variety of MMDEs. When checking a new MMDE it may be checked against the concept structures stored, for example, in the DB 150, and upon detection of a match providing the concept information about the MMDE. With the number of concept structures being significantly lower than the number of MMDEs the solution is cost effective and scalable for the purpose of identification of content of a MMDE.
As noted above the creation of a concept structure seeds from a set of initial training set of MMDEs that are either input to system or saved in a central repository. In some cases a user of the DCC system 100 may request for a concept structure that has not existed in the system 100 and there is no initial training set to create the requested concept structure. According to the disclosed embodiments, such a concept structure is created through a representation of known concept structures (hereinafter “sub-concepts”) existing in the DCC system 100 (hereinafter “sub-concepts”). The sub-concepts provide sparse representation of a (un-known) concept structure request to be created.
FIG. 7 shows an exemplary and non-limiting flowchart 700 describing a process for generating taxonomies within clusters in accordance with an embodiment. A taxonomy is related to a plurality of multimedia content data element (MMDEs) and generated using at least one MMDE. In S710, the at least one MMDE together with a request to generate taxonomies are received.
In an exemplary embodiment, a taxonomy may be a set of related concept structures. Two signatures representing concept structures are considered to be related if the two signatures sufficiently match. Signatures representing concept structures may be, e.g., signatures of MMDEs included in the content structures. In an embodiment, matching occurs as described in the '400 patent. If the two signatures overlap in more than a predetermined threshold level (e.g., 50% overlap), the two signatures are considered to be sufficiently matching. Concept structures may be considered to be related if a sufficient number or percentage of signatures of MMDEs included in one of the content structures are considered to be sufficiently matching. As a non-limiting example, if 40% of the signatures of MMDEs included in a first concept structure have a 60% overlap with signatures of MMDEs included in a second concept structure, the concept, the two concept structures may be considered to be related.
The request is received at the DCC system 110. The communication with the DCC system 100 is through, for example, a server or a client node being connected to the DDC system 100 a through a network, which may be the Internet, a LAN, and the like.
In S720, at least one signature for the received MMDE is generated. The signature(s) may be generated by the SG 120 as discussed in greater detail above. In S730, it is checked whether the generated signature matches one or more existing clusters and, if so, execution continues with S740; otherwise, execution continues with S750. In an embodiment, the generated signatures are matched to a SRC of each cluster maintained by the DCC 100. If the signatures generated for the MMDE and a respective SRC overlap in more than a first predetermined threshold level (e.g., 85% overlap), the MMDE is considered to fully match the concept structure of the respective SRC.
In S740, the at least one matching signature is associated with each matching cluster. In S750, inter-matching is performed between the at least one generated signature and a plurality of signatures previously generated by the DCC system 100. The inter-matching is performed as discussed in the '400 patent.
In S760, a clustering process takes place to generate a reduced cluster of the generated signatures and the matching signatures identified through the inter-matching process. In an embodiment, S760 is performed as discussed in the '400 patent. As may be necessary throughout these steps, the DB 150 or any other memory element may be used to store intermediate or final results. In S770, the at least one signature generated for the input MMDE is analyzed with respect to each SRC of the matching clusters produced at S760. In an embodiment this process is performed by the CG 140.
The output of S770 is at least one taxonomy related to received MMDEs and to one or more MMDEs represented by the clusters. In an embodiment, a taxonomy is considered to be related to a received MMDE if at least one of the signatures of the MMDE represented by the SRCs sufficiently matches the received MMDE. In an embodiment, matching is performed as described in the '400 patent. If the signature generated for the received MMDE and for a respective MMDE represented by the SRCs match in more than a predetermined threshold level (e.g., 50% of the signature representing the SRCs match the signature of the received MMDE), the taxonomy is considered to be related to the received MMDE.
In an embodiment, the taxonomy enables the determination of concept structures and determines their relations to other concept structures. For example, if one MMDE is matched to a plurality of MMDEs above a certain threshold, such MMDE is determined as a concept structure. The determination of concept structures is discussed further herein above with respect to FIG. 5.
In S780, it is checked whether there are additional MMDEs to be processed, and if so returns to S710; otherwise, execution terminates.
As a non-limiting example, an image of Kobe Bryant holding a basketball is received by the system 100 (the “Basketball image”). Signatures are generated respective of the Basketball image by the SG 120. The signatures are then matched to existing clusters in order to generate the taxonomy of the Basketball image with respect to other existing images. A match is found between the Basketball image and two other images: an image of the Los Angeles Lakers® (the team Kobe Bryant belongs to), and an image of Kobe Bryant wearing a suit. Based on the match type, the DCC system 100 determines the taxonomy between the Basketball image and the existing images. According to this example, as the image of the Los Angeles Lakers also matches a plurality of additional signatures above a certain predetermined threshold, the image of the Los Angeles Lakers is determined as a concept structure. As a result, other signatures that match this concept structure are determined to be concepts within this concept structure. A second match is identified between the Basketball image and the image of Kobe Bryant wearing a suit. As the signatures of the image of Kobe Bryant wearing a suit matched the signatures of the Basketball image, the image of Kobe Bryant wearing a suit is determined to be a concept that is related to the concept of the Basketball image.
Thus, the DCC system 100 enables the generation of taxonomies without any further inputs. The DCC system 100 is capable of automatically and, in an unsupervised fashion, creating concept structures of a wide variety of MMDEs. A new MMDE may be checked against the concept structures stored, for example, in the DB 150, and, upon detection of a match, concept information about the MMDE may be provided. With the number of concept structures being significantly lower than the number of MMDEs, the solution is cost effective and scalable for the purpose of identification of content of a MMDE.
The various embodiments disclosed herein may be implemented as hardware, firmware, software, or any combination thereof. Moreover, the software is preferably implemented as an application program tangibly embodied on a program storage unit or computer readable medium consisting of parts, or of certain devices and/or a combination of devices. The application program may be uploaded to, and executed by, a machine comprising any suitable architecture. Preferably, the machine is implemented on a computer platform having hardware such as one or more central processing units (“CPUs”), a memory, and input/output interfaces. The computer platform may also include an operating system and microinstruction code. The various processes and functions described herein may be either part of the microinstruction code or part of the application program, or any combination thereof, which may be executed by a CPU, whether or not such computer or processor is explicitly shown. In addition, various other peripheral units may be connected to the computer platform such as an additional data storage unit and a printing unit.
All examples and conditional language recited herein are intended for pedagogical purposes to aid the reader in understanding the principles of the invention and the concepts contributed by the inventor to furthering the art, and are to be construed as being without limitation to such specifically recited examples and conditions. Moreover, all statements herein reciting principles, aspects, and embodiments of the invention, as well as specific examples thereof, are intended to encompass both structural and functional equivalents thereof. Additionally, it is intended that such equivalents include both currently known equivalents as well as equivalents developed in the future, i.e., any elements developed that perform the same function, regardless of structure.

Claims

What is claimed is:

1. A method for generating a taxonomy of a plurality of multimedia data elements (MMDEs) based on at least one MMDE, comprising:

receiving the at least one MMDE;

generating at least one signature for the at least one received MMDE;

matching the at least one generated signature to a plurality of clusters to find at least one matching cluster;

associating the at least one generated signature with each of the at least one matching cluster;

analyzing the at least one generated signature with respect to a signature reduced cluster (SRC) of each of the at least one matching cluster to generate a taxonomy, wherein the taxonomy relates to the at least one received MMDE and an MMDE respective of each of the at least one matching cluster.

2. The method of claim 1, further comprising:

when at least one matching cluster is not found, performing an inter-matching process between the plurality of signatures stored in a database and the at least one generated signatures, wherein the plurality of signatures are generated for previously received MMDEs;

generating at least one cluster respective of the inter-matching results; and

analyzing the at least one generated signature with respect to a SRC of each the at least one generated cluster to produce the taxonomy.

3. The method of claim 2, wherein the method is performed by a DCC system.

4. The method of claim 3, further comprising:

saving the generated new data structure in a database communicatively connected to the deep-content-classification (DCC) system.

5. The method of claim 3, wherein the DCC system includes:

an attention processor (AP) for generating a plurality of items from the at least one MMDE;

a signature generator (SG) for generating at least one signature responsive to at least one MMDE; and

a concept generator (CG) for matching the at least one generated signature to a plurality of clusters to find at least one matching cluster.

6. The method of claim 1, wherein the at least one MMDE is at least one of: graphics, a video stream, a video clip, an audio stream, an audio clip, a video frame, a photograph, and images of signals.

7. The method of claim 6, wherein the images of signals are one of: medical signals, geophysical signals, subsonic signals, supersonic signals, electromagnetic signals, and infrared signals.

8. A non-transitory computer readable medium containing instructions that, when executed, perform the method of claim 1.

9. A system for generating concept structures, comprising:

an interface to a network for receiving a request to create a new concept structure, wherein the request includes at least a multimedia data element (MMDE) related to the new concept structure;

a processor; and

a memory connected to the processor, the memory contains instructions that when executed by the processor, configure the system to:

receive the at least one MMDE;

generate at least one signature for the at least one received MMDE;

match the at least one generated signature to a plurality of clusters to find at least one matching cluster;

associate the at least one generated signature with each of the at least one matching cluster; and

analyze the at least one generated signature with respect to a signature reduced cluster (SRC) of each of the at least one matching cluster to generate a taxonomy, wherein the taxonomy relates to the at least one received MMDE and an MMDE respective of each of the at least one matching cluster.

10. The system of claim 9, further configured to:

when at least one matching cluster is not found, perform an inter-matching process between the plurality of signatures stored in a database and the at least one generated signatures, wherein the plurality of signatures are generated for previously received MMDEs;

generate at least one cluster respective of the inter-matching results; and

analyze the at least one generated signature with respect to a SRC of each the at least one generated cluster to produce the taxonomy.

11. The system of claim 10, wherein the system further includes:

a signature generator (SG) for generating at least one signature responsive to at least one at least one MMDE; and

12. The system of claim 10, wherein the at least one MMDE is at least one of: graphics, a video stream, a video clip, an audio stream, an audio clip, a video frame, a photograph, and images of signals.

13. The system of claim 10, wherein the images of signals are one of: medical signals, geophysical signals, subsonic signals, supersonic signals, electromagnetic signals, and infrared signals.