116). If the extracted data features belong to the feature set the feature validation center 108 triggers feature-enabled network operations at least in part based on a set of valid data features, wherein the set of valid data features is associated with the feature set and the extracted data feature. The set of valid data features includes a list of feature IDs with respective feature similarity. The feature validation center 108 can trigger the feature-enabled network operations by sending the set of valid data features to a compression controller 110. Else, namely the extracted data feature does not belong to the feature set the feature validation center 108 generates and sends feedback regarding same to the feature extraction router 106 together with a request for feature re-extraction to the feature extraction router 106. This feedback can indicate information (e.g., words, phrases or other feature in the data) which at least in part describes potential data features that may be considered for data extraction. A potential data feature that includes the information may have a high probability of being selected as a data feature for subsequent extraction. This potential data feature may be associated with the data feature selected for subsequent extraction and existing data features already known or previously extracted. The greater the similarity between the existing data features and the data feature, the more likely the extracted data feature is included in the feedback as a potential data feature for extraction. Optionally, this extracted data feature can be considered as a potential data feature.

After receiving the feedback, the feature extraction router 106 extracts a new data feature with respect to the feedback received from the feature validation center 108. For example, the new data feature is identified and may include at least one potential data feature specified in the feedback from the feature validation center. The new data feature is subsequently sent to the feature validation center 108 for verification. If this new data feature passes the verification process, the feature validation center 108 obtains a set of valid data features which includes this new data feature. Then the feature validation center 108 triggers feature-enabled network operations at least in part based on the set of valid data features, the set of valid data features including the valid data feature which includes the new data feature. In this case, it may be reasonably appreciated that the system 100 could improve the accuracy of data feature extraction as a small level of similarity between the data feature and existing data features is achieved. As can be readily appreciated, this can be an iterative process in order to identify a data feature for extraction.

In some embodiments, the feature validation center 108 may check or evaluate the extracted data feature in order to determine if this extracted data feature is related to privacy information or information relating a person. As would be readily understood, it may be desired to note or identify sensitive/interesting data, or data features related to human beings as this type of information may require a level of protection. As such, selective encryption or combined encryption and compression may be used to protect sensitive/interesting data. In these instances, a privacy indication of whether sensitive/interesting data protection is required, the privacy indication is generated by the feature validation center 108, and is sent to the compression controller 110. In an optional manner, the privacy indication of whether sensitive data protection can be included in the set of valid data features together with the list of feature IDs and their respective feature similarity.

According to some embodiments, after receiving the list of feature IDs with their respective feature similarity and in some instances with the privacy indication, the compression controller 110 dynamically selects (or generates) a compression solution based on this input. This input may include the list of feature IDs with feature similarity, the privacy indication, compression performances on existing compression solutions (e.g., which may be available through an offline training function).

In some embodiments this input to the compression controller 110 may further include an indication from the data consumer 102. This indication from the data consumer 102 may indicate whether the decompressed data meets the necessary service requirements as previously defined by the request received from the data consumer 102.

According to embodiments, a parameter configuration function 114 generates and implements a parameter configuration including one or more parameters for the selected compression or compression and encryption of the extracted data feature. These parameters can be configured for the feature extraction router 106 and the data consumer 102. It may be important to note that the compression controller 110 and the parameter configuration function 114 may be deployed by the same provider. This is illustrated by the dashed box surrounding the compression controller and the parameter configuration function. However, it will be readily understood that the compression controller and the parameter configuration function may be provided by different entities, and thus suitable operative communication therebetween would be required.

In some embodiments, when the data consumer 102 accesses the required data, it may be necessary to check permissions regarding access of the required data by access control. In this instance, the feature validation center 108 can send the valid data feature with the data consumer ID (which can be associated with the data consumer 102, for example) to an access controller 112 for enabling accessing data control after feature validation.

FIG. 2 depicts a method 200 of feature extraction which may be implemented by the system 100 of FIG. 1. The method 200 includes, at step 202, receiving, by a feature extraction router, a request for feature extraction and at step 204, extracting, by the feature extraction router, a data feature (the data feature can also be referred to as extracted data feature), wherein in some embodiments, the data feature is at least in part identified based on feedback from a feature validation center. The method 200 further includes, at step 206, sending, by the feature extraction router, the extracted data feature to the feature validation center and at step 208, validating, by the feature validation center, the extracted data feature to obtain valid data features. The evaluation or assessment of the extracted data feature by the feature validation center can subsequently be feed back to the feature extraction router thereby in essence guiding data feature extraction by the feature extraction router. Specifically, the feature validation center determines whether the data feature belongs to a feature set, the feature set includes existing data features provided by a third party off-line (e.g. offline training function 116). Upon determination that the data feature does not belong to the feature set, the feature validation center generates feedback which indicates potential data features. The potential data features can be identified from or associated with the extracted data feature and the existing data feature in the feature set. The feature validation center can send the feedback to the feature extraction router in order to initiate the feature extraction router re-extracts a new data feature. Upon determination that the data feature belongs to the feature set, feature-enabled network operations are triggered, wherein the feature-enabled network operations are at least in part based on a set of valid data features, wherein the set of valid data features is associated with the feature set and the data features.

According to some embodiments, this feedback from the feature validation center can indicate information (e.g., words, phrases or other feature in the data) which describe aspects of a data feature that may be considered as a potential data feature for extraction. This information is generated by a feature validation center, wherein this information is indicative of a level of similarity between the extracted data feature and the existing data features already known or previously extracted. Feature similarity can be calculated by the feature validation center and can describe a level of similarity between the extracted data feature and the existing data feature (e.g., wherein an existing data feature can be a data feature identified during offline learning and wherein these existing data features can be provided by an offline training function). The greater the similarity between the extracted data feature and the existing data feature in the feature set, the more likely the extracted data feature is included in the feedback as a potential data feature for extraction.

In some embodiments, the set of valid data features includes a list of feature IDs with their own feature similarities. The set of valid data features further comprises a privacy indication which is generated by the feature validation center based on the data features. The privacy indication indicates of whether sensitive/interesting data protection is required for feature-enabled network operations.

In some embodiments, the method 200 further includes, at step 210, sending, by the feature validation center, the set of valid data features to a compression controller. The method 200 further includes, at step 212, selecting, by the compression controller, a compression solution based on an input, and at step 214, sending, by the compression control, the compression solution to a parameter configuration function. The method 200 further includes, at step 216, generating, by the parameter configuration function, a parameters configuration based on the compression solution. In some embodiments, the input may comprise the set of valid data features, compression performances on existing compression solutions, an indication which indicates whether the decompressed data meets the necessary service requirements as previously defined by the request received from a data consumer.

Currently, it is known that typically data feature extraction is performed by extracting data features for example simply based on frequency of occurrence, without consideration of the usefulness of the data features. In some instances, these extracted data features may not be useful. As a result, poor performance of network operations (e.g., feature based access control, or data compression) can occur. In contrast to known systems, the system of the present disclosure first extracts a data feature and subsequently validates the extracted data feature, for subsequently triggering proper or appropriate network operations. According to embodiments, after extracting the data feature, whether the data feature belong to a feature set will be determined, when the data feature does not belong to the feature set, feedback will be generated, it is the feedback from the feature validation center to the feature extraction router that can improve the accuracy or usefulness of a data feature being extracted. This interaction between the feature extraction router and the feature validation center can be an iterative process in order to improve upon one or more extracted data features enabling improved network operations, for example data compression.

According to embodiments of the present disclosure provides a method for feature validation. With reference to the system as illustrated in FIG. 1, the feature extraction router 106 extracts a data feature that is subsequently sent to the feature validation center 108. The process performed by the feature validation center includes determining whether the extracted data feature belongs to a feature set or not. This feature set can include one or more existing or previously extracted data features which occurred during operation of the system and/or may include existing data features that are identified during off-line learning.

FIG. 3 provides a flowchart 300 for feature validation. It may be helpful to note that an offline training function, for example as illustrated in FIG. 1 can be used to develop or create the feature set, wherein the feature set includes one or more data features, wherein each data feature can be associated with a feature ID and feature content.

Referring now to FIG. 3, the extracted data features 302 from the feature extraction router, for example feature extraction router 106 of FIG. 1, will be identified or evaluated in order to determine, at step 304, whether they belong to a feature set or not, through comparisons of feature similarity values with a threshold. Data features from this feature set may be provided by the offline training function 116 of FIG. 1 or may be identified during operation of the system and deemed to be suitable for inclusion in the feature set. Subsequently two options then follow upon the evaluation of the comparison.

In some embodiments, the feature validation center 108 calculates the level of data feature similarity between the data features 302 and one or more of the data features included in the feature set, at step 304. This feature set can include data features provided by the offline training function 116, for example. During the determination of similarity, feature validation center checks whether the feature similarity between the extracted data features 302 and the feature set exceeds a threshold at step 306. If the threshold is exceeded, the feature validation center 108 sends an indication of feature similarity together with the feature ID that can be associated with the particular data feature associated with the feature set, to the compression controller 110. Based on this information, the compression controller 110 subsequently generates a compression policy at step 308. If the level of similarity does not exceed the threshold, the feature validation center 108 sends a feedback to the feature extraction router 106 for the continuation of data feature extraction at step 310.

In some embodiments, a knowledge-based method or other similarity estimation method may be used to calculate feature similarity. For example, methods for similarity estimation can be associated with the use scatter plots, statistical techniques and piecewise linear interpolation. In some embodiments, subjective Bayes method to model concept semantic similarity and obtain semantic similarity of related words can be used. It will be readily understood that other concepts of similarity estimation can be used and would be readily understood by a worker skilled in the art.

In some embodiments, the feature validation center 108 checks the extracted data feature to determine whether the extracted data feature is related to privacy information or information of a natural person (i.e., a human being). It may be important to note that it is typically desired that sensitive or personal data associated with data features and data features related to human beings should be protected. As such, selective encryption or encryption joint compression or other encryption methods may be used to protect such sensitive or personal data.

FIG. 4 depicts a method 400 for data feature validation, at least in part based on the method illustrated in FIG. 3. The method 400 includes, at step 402, receiving a data feature. The method 400 further includes, at step 404, validating the extracted data feature and at step 406, generating a privacy indication and obtain a set of valid data features for the validated data feature.

Benefits of embodiments of the system and method for data feature extraction as defined in the present disclosure include consideration of feedback from the network during subsequent extraction of data features. In this manner, the extracted features can be tuned or refined, for example based on similarities to previously extracted or known data features. This is a benefit over the prior art which does not consider any type of feedback. Further, the feedback indicates information (e.g., words, phrase) describing the potential data features, wherein this information can be generated based on feature similarity. Feature similarity can describe a level of similarity of an extracted data feature and an existing data feature (which in some instances may be identified during offline learning of data features).

According to an embodiment of the present disclosure there is provided a method for data compression. For example, with reference to the system of FIG. 1, the compression controller 110 can dynamically select or generate a compression solution based on input received from the feature validation center. This input can include a list of feature IDs with a level of feature similarity, a privacy indication, compression performance relative to existing compression solutions (which may be defined by offline training samples), and an indication from data consumer 102. The indication from data consumer 102 may indicate, upon decompression of the data transmitted to the data consumer, whether the data meets the necessary service requirements of the data consumer 102. One or more of these requirements may be provided by the data consumer in an initial service request.

Embodiments relating to the generation of a compression solution will now be discussed. Referring now to FIG. 5 there is depicted a flowchart for data compression, according to embodiments. If a compression solution 502, which is obtained from offline training samples, is matched to the received feature ID, wherein for example the data feature associated with the received feature ID has been previously compressed using the compression solution 502, the compression controller 110 selects a compression solution or policy 506 based on the received feature ID and sends the selected compression ID associated with that compression solution to the parameter configuration function 114. The parameter configuration function 114 subsequently performs parameter generation 508 and determines compression parameter configurations 510a and 510b. It is noted that in some embodiments, the parameter configuration function 114 can be merged into the compression controller 110.

In some embodiments, if the privacy indication defines that data associated with the feature ID needs privacy protection, the compression controller 110 also sends an ID of an encryption algorithm to the parameter configuration function 114 which enables the determination or evaluation of the encryption parameter configuration by the compression controller.

In some embodiments, the compression controller 110 may send partial parameters compression solution or policy to the parameter configuration function 114. The parameter configuration function 114 may generate parameters 508 for compression and if required encryption, and subsequently sends these generated parameters to the feature extraction router 106 in the form of configuration 510a and the data consumer in the form of configuration 510b. Upon received of the respective configurations 510a and 510b, the feature extraction router 106 and data consumer 102 configure the compression solution based on the received configurations.

If a compression solution determined from offline training samples is not matched to the received feature ID, the compression controller 110 determines a compression solution based on at least one of an input, compression performances of existing compression solutions and an indication from a data consumer. The input can include the set of valid data features and the indication can be indicative of whether, upon decompression, data meets the necessary service requirements. The compression controller 110 subsequently generates a preferred compression solution based on the feature similarity, the private indication, compression performances on existing compression solutions (e.g., offline training samples), and the indication from the data consumer 102. Upon the generation of the preferred compression solution a compression ID is assigned and the compression controller 110 sends the compression ID to the parameter configuration function 114 for the determination or evaluation of compression parameter configurations 510a and 510b. The compression controller 110 may also send an ID of an encryption algorithm to the parameter configuration function 114 for the determination of an encryption parameter configuration in instances where the privacy indication indicates that the required data requires privacy protection. In some embodiments, the compression controller 110 may send partial parameters compression solution or policy to the parameter configuration function 114. The parameter configuration function 114 may generate parameters 508 for compression and if required encryption, and subsequently sends these generated parameters to the feature extraction router 106 in the form of configuration 510a and the data consumer in the form of configuration 510b. Upon received of the respective configurations 510a and 510b, the feature extraction router 106 and data consumer 102 configure the compression solution based on the received configurations.

Currently, considerations for data compression are based on the extracted feature, and as such a current compression controller obtains sensitive features which may result in the leaking of privacy of the data owner or the data consumer. As such, benefits of embodiments of the system and method for data feature compression as defined in the present disclosure include consideration relating to data compression based on a list of feature similarities, indications and compression performances on existing compression solutions, in order to better protect privacy. Moreover, embodiments may provide compression solutions for any features (including matched features and unmatched features). Unmatched features can be used to define extracted data features are not included in existing data features. Similarly, matched features means that extracted data features are included in existing data features.

According to embodiments, a procedure for data compression is provided, wherein when the data consumer 102 requests a data compression service. The network first checks whether the data consumer 102 has a permission to obtain the requested data and upon authorization, the network implements data compression to reduce the network bandwidth and communication overhead. However, due to environments wherein elements associated with a system can lack trust therebetween, for example in instances wherein the elements of the system are provided by multiple non-trusted parties or where the data consumer is untrusted with private data, it is desirous that sensitive data or personal data relating to an activity performed by an element of the system be protected. In addition, sensitive data should be protected during transmission and thus transmission can be enabled using a level of encryption.

FIG. 6 depicts a procedure 600 for data compression according to embodiments. In the procedure 600, the data consumer 102 sends a request 602 to the feature extraction router 106 through for example the service controller 104 (of FIG. 1). The request 602 can include information relating to the data consumer, for example a data consumer ID, service information indicating what data can be sent to data consumer 102, and one or more security requirements if security of the information being sent to the data consumer is necessary. In some embodiments, the request 602 may be associated with multi-view live video data, and the request 602 may also include a natural person's interactive activity that may be able to provide the data consumer 102 with information indicative of gestures, body positions and/or angles of view in the video/data.

After receiving the request 602, the feature extraction router 106 extracts a data feature (data feature extraction 604) based on the service information and the human's interactive activity. The feature extraction router 106 then sends a feature validation request 606 which includes the data feature and if required one or more security requirements, to the feature validation center 108.

The feature validation center 108 implements feature validation to obtain a set of valid data features, based on a feature set representative of known or previously extracted data features, which may be provided by an offline training function 116. The feature set can include one or more data features and an associated feature ID for each of the one or more data features.

In some embodiments, if the extracted data feature belongs to a feature set, the feature validation center 108 sends a feature validation response 610 to the feature extraction router 106, wherein the feature validation response can be indicative of the match.

In some embodiments, the feature validation center 108 calculates the level of data feature similarity 608 between the extract data feature and one or more of the data features included in the feature set. This feature set can include data features provided by the offline training function 116, for example. The feature validation center 108 can be configured to check the level of feature similarity and determine whether this level of feature similarity exceeds a threshold. For example, the threshold can be configured as a fixed threshold or a dynamic threshold which may be defined at least in part based on extracted data feature characteristics or one or more other parameters. Upon determination of the level of feature of similarity, the feature validation center sends an appropriate data feature validation response 610 to the feature extraction router 106.

In some embodiments, if the extracted data feature belongs to a feature set or if the level of feature similarity exceeds the threshold, the feature validation response 610 indicates that the extracted data feature is accepted. However, in instances where the extracted data feature neither belongs to the feature set nor the level of similarity exceeds the threshold, the feature validation response can include feedback information relating to the extracted data feature. For example, the feedback information can indicate information (e.g., words, phrases or other data features) that can describe characteristics of suitable potential data features that may be extracted. This feedback information can be generated by the feature validation center 108 based on the data feature similarity previously evaluated. For example, the higher the feature similarity of the extracted data feature, the more likely the data feature is included in the information feedback as a potential data feature.

Following the feature validation center 108 sending a feature validation response 610 to the feature extraction router 106, if the feature validation response 610 includes the feedback information, the feature extraction router 108 extracts a new data feature at least in part based on the feedback information.

According to embodiments, if the feature validation response 610 indicates that the extracted data feature is accepted, the feature validation center 108 sends an access request 612 to access controller 112. This access request 612 includes the data consumer ID, and the extracted data feature.

According to embodiments, the access controller 112 can implement access control based on the extracted data feature and send back an access response 614 to the feature validation center 108. The access response 614 can indicate the data consumer's 102 permission with respect to the requested data, which may at least in part be based on the information received in the initial service request. If the access response 614 indicates that the data consumer 102 does not have permission to obtain the requested data, the feature validation center 108 rejects the service request, and sends back a service response 616 to the data consumer 102. This response 616 indicates that the data consumer 102 does not have permission to obtain the requested service.

According to embodiments, if the access response 614 indicates that the data consumer 102 has permission, the feature validation center 108 may check the extracted data feature in order to determine if the extracted data feature includes any information that may be deemed to be private or the privacy thereof is to be protected. For example, the feature validation center can determine if the extracted data feature is related to a human being's privacy information or a human being's interesting information. If privacy of some of the information associated with the extracted data feature is desired or required, selective encryption or encryption joint compression may be used to at least protect private data associated with the extracted data feature. The feature validation center 108 then sends a compression request 618 to the compression controller 110. This compression request 618 may include the set of valid data features, a list of feature IDs with the corresponding level of feature similarity, a privacy indication indicating whether data compression requires encryption or other form of privacy protection prior to being transmitted to the data consumer.

In some embodiments, in instances where an existing compression solution is known (e.g., for example a compression solution determined or configured based on offline training samples) which is matched to the received feature ID, the compression controller 110 sends a parameter request 624 to the parameter configuration function 114. This parameter request may include compression ID, and partial parameters about compression. This parameter request may also include an encryption ID if the privacy indication shows data compression requires at least partial data protection. In some embodiments, offline training function samples may include information including feature ID, compression ID with a corresponding compression performance associated with the compression method associated with the compression ID.

In some embodiments, where a known compression solution is unavailable, the compression controller 110 is configured to determine a compression solution based on at least one of an input, compression performances of existing compression solutions and an indication from a data consumer. The input can include the set of valid data features and the indication can be indicative of whether, upon decompression, data meets the necessary service requirements. The compression controller