Content-based storage management

FIELD OF THE INVENTION

0001The present invention relates to a system and a method for content-based storage management, and in particular, for such a system and method in which decisions concerning the location and/or timing of storage of data are based upon the content of the data.

BACKGROUND OF THE INVENTION

0002Storage facilities for digital information are a critical resource. The demand for storage space for both conventional data, such as text documents and other human readable files, and multimedia streams, such as audio and/or video data, has increased significantly. Such an increase results from a number of different factors, such as legal requirements to store and maintain certain types of information; an increase in the different types of data which are being stored; and even an increase in the size of individual units of data, such as word processing document files, video data files and so forth. This increased demand has in turn resulted in a higher demand for storage space, and in particular for storage space which is accessible "on-line.

0003As the demand for on-line storage space increases, a number of options are possible to fulfill that demand. For example, additional hardware, such as magnetic media devices ("hard disk drives"), may be purchased to increase the available amount of electronically accessible storage space. However, as the quantity of such hardware devices increases, the management problem for electronic management of these devices also increases. Furthermore, merely increasing the storage space may be both wasteful and unnecessary, since not all of the data may be required, or at least not required for immediate access.

0004The problem may be partially alleviated through the use of a mixture of different types of storage facilities. For example, on-line storage refers to direct-access, permanently mounted storage areas, such as magnetic (or other types of media) disk drives and disk arrays. The time required for access to such storage areas to be made is typically measured in fractions of a second. Since not all data may need to be stored in on-line storage, which is fast but also expensive, near-line storage is also available. Near-line storage is based upon an automatically (machine) operated storage area, such as optical disks residing on a disk "jukebox" or tapes in an automatic tape library. Such automatically operated storage devices are able to store and automatically access a relatively large amount of data with fewer physical reading devices, or drives, for reading the data. This type of storage is less expensive, but also somewhat slower for accessing the data, such that access times are measured in seconds to minutes, or even longer, depending upon the availability of physical drives for reading the storage media. On-line or near-line storage may also feature a system with a plurality of physical drives, connected together, for example in a LAN (local area network) or WAN (wide area network).

0005Off-line storage is the least expensive type of storage, but is also the slowest for access, as it does not permit automatic electronic access. Instead, manual operation of the storage devices and physical drives is required by a human operator. The number of physical drives is greatly reduced compared to the number of storage devices (or at least the amount of available storage space). However, the access time for data from such devices is measured from minutes to hours, depending upon the availability of the human operator and the location of the storage devices, as well as the availability of the physical drives.

0006Other types of storage devices and functions may also be used, in addition to, or in replacement for, the above-described devices and functions. In any case, the difficulty with a mixed system, or a system in which different types of storage areas (topology) are used, with different types of storage devices and different accessibility (particularly with regard to access time), is the management of the data. Certain types of data may be more important, or at least more time-critical for access, such that the access time may be very important for some types of data, and much less important for other types of data. Cost is also an important factor. Also, decisions must be made concerning the number and type of storage devices to be purchased, along with any required supporting devices and/or system support, such as human operators for example. Currently, these systems are designed and constructed manually, and decisions are made on the basis of some type of policy. However, the operation of the actual system and even the design itself may not be optimal for a particular organization.

0007<patcit id="pcit0001" dnum="EP0950965A"><text>EP-A-0950965</text></patcit> relates to a document storage management system and method that manages the storage of documents based upon the similarity of the content of the documents. Groups of documents are created based upon the similarity of the contents of the documents. The storage of the selected group or document is then managed by, for example, deleting, compressing, or copying.

0008The document of<nplcit id="ncit0001" npl-type="s"><text> Hiroshi Ishikawa et al.: A Document Warehouse: A Multimedia Database Approach. DEXA Workshop 1998: pages 90-94</text></nplcit> relates to a document warehouse as a software framework where multimedia documents are analyzed and managed for corporate-wide information sharing and reuse like a data warehouse for structured data. It describes a prototype document warehouse system, which supports management of simple and compound documents, keyword-based and content-based retrieval, rule-based classification, SOM-based clustering, and business rules.

0009<patcit id="pcit0002" dnum="EP0341230A"><text>EP-A-0341230</text></patcit> relates to a method for managing data in a data storage hierarchy, and a data storage hierarchy for implementing such method. The data storage hierarchy includes a data catalog and a plurality of data storage levels. The data catalog stores information indicating data management rules for the data stored in the hierarchy. Data stored in the hierarchy are flagged upon being referenced by a host processor. The flagged data is managed according to the management rules during periods of low hierarchy activity.

0010<patcit id="pcit0003" dnum="US5131087A"><text>US-A-5131087</text></patcit> relates to an apparatus including a number of data bases which are used by the expert system software to manage the computer system data storage devices. One element provided in this apparatus is a set of data storage device configuration data that provides a description of the various data storage devices and their interconnection in the computer system. A second element is a knowledge data base that includes a set of functional rules that describe the data storage device management function. These rules indicate the operational characteristics of the various data storage devices.

0011<patcit id="pcit0004" dnum="US5506986A"><text>US-A-5506986</text></patcit> relates to a media management system using a rule based system to manage the media assignment function for a data processor. This apparatus comprises a set of software programs that communicate in real time with various standard software components on the data processor and also retrieves information describing data stored on the data storage devices themselves to compile information concerning data set activity.

SUMMARY OF THE INVENTION

0012The background art does not teach or suggest a solution to the problem of efficiently managing data storage. The background art also does not teach or suggest a solution to the problem of managing data storage for both cost efficiency and for suitable access times. The background art also does not provide a solution for storing data according to the content of the data, such that important data can be stored in a more accessible location/type of file storage. In addition, the background art does not teach or suggest a system and method for managing data, such that data is correctly stored, migrated and/or deleted, according to the content thereof.

0013The present invention overcomes these problems of the background art by providing a system and method for data management according to the content of the video or audio data. The present invention enables data to be stored in one of a plurality of different storage options according to at least one characteristic of the data, in which the at least one characteristic is related to the content of the data. It should be noted that at least one storage option according to the present invention is optionally deletion and/or destruction of the data, such that the data may optionally be removed from storage media or may optionally not be stored initially on the storage media. Therefore, a "storage option" according to the present invention includes any type of storage media, device, system or combination thereof, or deletion (removal) of the data.

0014Optionally and more preferably, the data is stored for a time interval according to the at least one characteristic of the data. Most preferably, the data is moved to a different type of storage option after an event has occurred, for example after the time interval has elapsed. It should be noted that movement or migration of the data may also include deletion or removal of the data.

BRIEF DESCRIPTION OF THE DRAWINGS

0015The invention is herein described, by way of example only, with reference to the accompanying drawings, wherein: <ul id="ul0001" list-style="none" compact="compact"><li><figref idref="f0001">FIG. 1</figref> is a schematic block diagram of an exemplary system and flow of operations according to the present invention;</li><li><figref idref="f0002">FIG. 2</figref> is a schematic block diagram of another exemplary system and flow of operations according to the present invention; and</li><li><figref idref="f0003">FIG. 3</figref> is a schematic block diagram of a detailed exemplary system according to the present invention.</li></ul>

DESCRIPTION OF THE PREFERRED

EMBODIMENTS

0016The present invention is of a system and method for data management according to the content of the video or audio data. The present invention enables data to be stored in one of a plurality of different storage options according to at least one characteristic of the data, in which the at least one characteristic is related to the content of the data. It should be noted that at least one storage option according to the present invention is optionally deletion and/or destruction of the data, such that the data may optionally be removed from storage media or may optionally not be stored initially on the storage media. Optionally and more preferably, the data is stored for a time interval according to the at least one characteristic of the data. Most preferably, the data is moved to a different type of storage option after an event has occurred, for example after the time interval has elapsed. It should be noted that movement or migration of the data may also include deletion or removal of the data.

0017According to preferred embodiments of the present invention, the at least one characteristic of the data, according to which the storage option is selected, is examined by a rule engine. Preferably, the rule engine compares the at least one characteristic of the data to at least one rule, and then selects the storage option (or options) according to that rule. The rule engine therefore more preferably operates as a filter, for determining which storage option(s) is most appropriate for the examined data. The storage decision is then preferably implemented by a storage manager.

0018The rules according to which the rule engine operates are optionally manually entered by a human operator, or alternatively may optionally be generated automatically according to a predefined business rule or according to an automatically generated business rule, or a combination thereof.

0019Preferably, the present invention is operative with a system featuring a plurality of different storage options. More preferably, these different storage options include at least two different storage options having different types of accessibility. Examples of storage options having different types of accessibility include but are not limited to on-line storage, near-line storage and off-line storage. The type of storage media which is used for any particular storage option is not limited according to the present invention, as the present invention is operable with any suitable type of storage media, including but not limited to DAT (tape-based storage), AIT (also tape-based storage), magnetic storage media, optical disks, CD-ROM or a mass storage device of any type, or any type of storage system, or any combination thereof.

0020The at least one characteristic of the data, which is related to the content of the data, may be obtained in a number of different ways. For example, the data may optionally and preferably have associated metadata, which is related to the content of the data. The metadata is more preferably added through annotation of the data itself. Such annotation is performed automatically. More preferably, automatic annotation is performed after the data is automatically analyzed. The associated metadata is then preferably used to determine which storage option should be used for the data, and more preferably also the time interval during which the data should be placed in that storage option.

0021As previously described, the data is more preferably filtered by a rules engine, according to at least one characteristic of the data. For the implementation of the present invention with metadata, the filtering process is more preferably performed according to the associated metadata.

0022Automatic analysis of the data is more preferably performed according to the type of data being analyzed. Examples of different types of automatic analysis processes which may optionally be performed include speech-to-text conversion for voice communication data, a video analyzer for video data, and analyzers for user interface data. These different types of data analysis processes are preferably performed according to the source of data. Examples of different types of data sources include at least video data or audio data (including also voice communication data such as voice over IP (VoIP) data, streaming audio data and any other type of audio-related data), and user interface data.

0023In addition, the present invention is useful for the collection of data about a user interface function. Examples of such user interface functions include at least information provided through an audible interface such as a synthesized voice; or information provided through the display of video data.

0024According to preferred embodiments of the present invention, the data is preferably "migrated" or moved from a first storage option to a second, different storage option after a time interval has expired. The time interval is preferably determined according to the metadata.

0025The principles and operation of the method according to the present invention may be better understood with reference to the drawings and the accompanying description.

0026Referring now to the drawings, <figref idref="f0001">Figure 1</figref> shows a first exemplary system 10 for managing data according to the content of the data, with regard to the flow of operations through system 10. As shown, system 10 features at least one input data source 12. Examples of different types of input data sources 12 include at least video data or audio data (including also voice communication data such as voice over IP (VoIP) data, streaming audio data and any other type of audio-related data) and user interface data. With regard to user interface data, in which the action of a user upon operating a computational device and/or a peripheral device and/or input device thereof causes the data to be generated, optionally and preferably the data is in the form of an event. Each action of the user preferably causes an event to be generated. The event may then optionally form the data to be captured.

0027The captured data is passed to a format analyzer <b>14</b> for rendering the captured data into a common format for analysis. Format analyzer <b>14</b> preferably features a plurality of format modules <b>16,</b> each of which is suitable for data of a different type of format. For example, if the input data is voice communication data, then preferably a format module <b>16</b> converts the voice communication to textual data, for speech-to-text conversion. Different format modules <b>16</b> preferably handle other types of input data, as explained in greater detail with regard to <figref idref="f0003">Figure 3</figref> below.

0028According to a preferred embodiment of the present invention as claimed in claim 1, the common data format is optionally and preferably textual data. For this preferred embodiment, textual data is optionally not further preprocessed by a format module 16, or alternatively is only minimally processed. Other types of data in different data formats are then preferably converted to textual data by format module 16, as described with regard to <figref idref="f0003">Figure 3</figref> below.

0029Next an analysis module <b>18</b> preferably analyzes the data, once the data is in the common format. It should be noted that optionally analysis module <b>18</b> is able to handle a plurality of different data formats, through a single module or alternatively from a set of such modules (not shown). Preferably, analysis module <b>18</b> operates on a plurality of different types and/or sources of data simultaneously, for example as a multi-thread application. For the non-limiting, illustrative example, as described above, textual data is analyzed, as is well known in the art. One non-limiting example of a text analyzer software is the IntelligentMiner™ product (IBM Corp, see http://www-4.ibm.com/software/data/iminer/fortext/tatools.html as of December 31, 2001 for further details). This product is useful for analyzing text for a number of functions, such as locating information related to a topic, categorization of information and classification of information. Text analyzer tools are generally known in the art for extracting content and/or information related to the subject matter of text, for example according to one or more keywords, concepts or any other type of organization and/or analysis scheme.

0030According to another preferred embodiment of the present invention, the uniform format features a uniform data structure, with a plurality of different types and/or categories of information, for example data from screen events and voice data combined in a single file. This uniform data structure preferably is able to contain the different types or categories of characteristics which are of interest for being associated with the data, in order to determine the content of the captured data. Non-limiting examples of a uniform data structure which may optionally be implemented according to the present invention include structures which use XML (extensible mark-up language) or ASF (Advanced Streaming Format, from Microsoft Corp., USA).

0031Analysis module <b>18</b> preferably extracts and/or creates, or otherwise determines, at least one characteristic of the captured data, preferably obtained in the uniform format. Optionally and more preferably, analysis module <b>18</b> obtains the at least one characteristic from the captured data in the form of metadata. This metadata is then optionally stored in a metadata database (not shown, see <figref idref="f0003">Figure 3</figref>).

0032According to an optional but preferred embodiment of the present invention, analysis module <b>18</b> also gives feedback for improving the performance of format analyzer <b>14</b> and/or format module <b>16,</b> in order to improve the operation of these components.

0033The captured data, with the at least one characteristic related to the content of data, which is optionally and more preferably in the form of metadata, is then passed to a rule engine <b>20</b>. Alternatively, rule engine <b>20</b> only receives the at least one characteristic, more preferably in the form of metadata. The captured data is then passed directly to a storage manager <b>22.</b> Rule engine <b>20</b> more preferably compares the metadata to at least one rule, which is most preferably a business rule specified by a manual human user, or alternatively may optionally be generated automatically according to a predefined business rule or according to an automatically generated business rule, or a combination thereof. Optionally, one or more rules may be fed to rule engine <b>20</b> through an interface <b>24</b>, such as a GUI (graphical user interface) for example. As a non-limiting example, interface <b>24</b> may optionally be a simple Web browser-based interface. Rule engine <b>20</b> then preferably determines the type of storage option (or options) according to one or more rules, as selected through the comparison of the metadata (or characteristic) of the captured data to the rule(s). Additionally or alternatively, the output of rule engine <b>20</b> is optionally and preferably fed back to format analyzer <b>14,</b> and/or format module <b>16,</b> and/or analysis module <b>18.</b>

0034Rule engine <b>20</b> optionally and more preferably determines both the type of storage option (or options), which should be selected for the particular captured data, and also the term of storage. Most preferably, the captured data is initially stored with a first storage option, and then is migrated (moved) to at least one additional storage option after a period of time has elapsed. This period of time is also most preferably determined according to at least one rule by rule engine <b>20.</b>

0035Rule engine <b>20</b> may optionally perform an action according to a rule and/or event, in which the event may optionally and preferably trigger automatic application of the rule. Examples for actions based on rules are given below. One example of actions which are based on rules include but are not limited to manipulations of stored data. For example, the compression of the data may optionally be altered after an event has occurred, such as a period of time has elapsed. A non limiting example of a reason for altering such compression is to enable more rapid playback of the data.

0036Previously stored data may optionally and preferably be updated with business data. Non limiting examples of such data include the addition of social security or identification number, customer identifier, preferred customer status information and so forth.

0037Current transaction data may also optionally and preferably be linked to the previously stored data file. Previously stored data may also optionally be updated. A non limiting example of such linking may optionally be performed by linking transactions performed by a certain high-status or preferred customer to past transaction by that customer.

0038Another event/action example may optionally be performed with multiple mirroring of data, for example by distributing identical data to several destinations and/or pre-defined locations and/or storage options. Such mirroring may optionally be performed for redundancy purposes, for example for security of the stored data, by duplicating to multiple storage locations/options, and/or for general availability reasons.

0039Storage manager <b>22</b> preferably then receives the output of rule engine <b>20</b> and the captured data. The output of rule engine <b>20</b> preferably includes at least one storage option for the captured data. As previously noted, this storage option could be a type of storage media or deletion and/or removal and/or destruction of the captured data. More preferably, the storage option includes a particular storage device <b>26</b> into which the captured data should be placed. Alternatively, storage manager <b>22</b> could determine the identity of the particular storage device <b>26</b> for storing the data.

0040<figref idref="f0002">Figure 2</figref> shows a different flow arrangement of the system of <figref idref="f0001">Figure 1</figref>, as another example of a system according to the present invention. In this arrangement, system <b>10</b> again features a rule engine <b>40.</b> However, in this implementation, rule engine <b>40</b> is the initiating process or component for subsequent actions which are performed by system <b>10.</b> As show, rule engine <b>40</b> takes the input source according to metadata obtained from an analysis module <b>38.</b> Rule engine <b>40</b> then preferably sends the captured data, or alternatively only selected captured data, to a storage manager <b>42.</b> Storage manager <b>42</b> sends the captured data to the correct storage option, shown as preferably being a selected storage device <b>44,</b> according to a request for action by rule engine <b>40.</b> Optionally and more preferably, and most preferably as necessary, rule engine <b>40</b> feeds back the captured data, and/or information about the captured data, into the input sources. Improved metadata may optionally be obtained from analysis module <b>38.</b>

0041According to an optional implementation of the present invention, the user defines a task in rule engine <b>40</b> for archiving certain types of information, such as information about specific telephone calls. Rule engine <b>40</b> then preferably uses analysis module <b>38</b> to select specific data. Analysis module <b>38</b> may optionally be implemented as a call management server, for example. The selected specific data may be any one or more of voice data, data captured from user interface actions, video data VoIP. The data is obtained from input sources <b>36,</b> which may optionally be implemented as an input sources logger. The captured data is then transferred into a storage manager <b>42</b>.

0042This data can optionally be retrieved as required from storage devices <b>44,</b> more preferably directly by using rule engine <b>40</b> and/or storage manager <b>42.</b> Such retrieved data may then optionally and more preferably be fed into input sources <b>36</b> in order for the retrieved data to be widely accessible (available), and/or for further manipulation.

0043<figref idref="f0003">Figure 3</figref> shows an exemplary, detailed implementation of a system according to the present invention, which is related to the flow of operations shown in <figref idref="f0001">Figure 1</figref>. As shown, a system <b>50</b> preferably features a plurality of input data sources <b>52.</b> Input data sources <b>52</b> are shown as including a video source <b>54</b> for video data, an audio source <b>56</b> for audio data, a messaging source <b>58</b> for e-mail messages (including attachments), instant messaging and/or chat data, and a facsimile source <b>60</b> for data which is transmitted by facsimile.

0044The input data from input data sources <b>52</b> is preferably then fed to at least one format analyzer <b>62</b> for rendering the captured data into a common format for analysis. Format analyzer <b>62</b> preferably features a plurality of format modules <b>74,</b> each of which is suitable for data of a different type of format. For the purposes of illustration only and without any intention of being limiting, format modules <b>74</b> are shown as optionally and preferably including a video analyzer <b>76,</b> a text analyzer <b>78,</b> an audio analyzer <b>79</b> and an OCR (optical character recognition) module <b>80.</b>

0045An example of text analyzer <b>78</b> was previously described with regard to <figref idref="f0001">Figure 1</figref>. OCR module 80 may optionally be implemented as is well known in the art, for example through the use of OCR software having an algorithm which could easily be selected by one of ordinary skill in the art.

0046Video analysis may optionally be performed by video analyzer <b>76</b> as follows. Video data is obtained, for example from a camera as a non-limiting example of video source <b>54.</b> A frame-grabber is then preferably used to obtain at least one frame from the video data. The frame is preferably analyzed. More preferably, only a portion of the frame is stored as captured data. For example, if a video camera is used to monitor the entrance to a secure area, then optionally only those frames, or alternatively those portions of each frame, which feature a human subject near the actual entrance are of interest. Additionally or alternatively, changes in the background of each frame may optionally be detected and tracked, as being of interest.

0047One example of a type of analysis which may be performed with the video data is a motion detection algorithm, which is well known in the art. Another example is face recognition algorithms, which are also well known in the art. Non-limiting examples of video analysis algorithms are described at <u>http://www.cs.rochester. edu/u/nelson/research/motion/motion.html</u> as of December 31, 2001, for motion detection algorithms and at <u>http://wwwwhite.media.mit.edu/vismod/demos/facerec</u> as of December 31, 2001, for face recognition algorithms. More preferably, such analyses are performed with firmware, such as a DSP (digital signal processor) for example. The results may then optionally be stored as the captured data.

0048The output of format analyzer <b>62</b> preferably features at least one characteristic of the captured data which more preferably is metadata, as previously described. The metadata is more preferably stored in a metadata database <b>82.</b> A rule engine <b>84</b> then preferably analyzes the metadata from metadata database <b>82</b> (or alternatively obtained directly from format analyzer <b>62</b>), in order to apply one or more rules to the captured data. Rule engine <b>84</b> may optionally be implemented with the BlazeSoftware Advisor product of Blaze Software (see <u>http://www.blazesoft.com/products</u> as of December 31, 2001 for details). Rule engine <b>84</b> may also optionally be implemented as a rule/task engine from Nice Systems Ltd. (Ra'anana, Israel), for example based on business data. Regardless of the specific implementation, rule engine <b>84</b> preferably operates according to at least one business rule.

0049Rule engine <b>84</b> preferably compares the at least one characteristic of the data to at least one rule, and then selects the storage option (or options) according to that rule. Rule engine <b>84</b> therefore more preferably operates as a filter, for determining which storage option(s) is most appropriate for the examined data. The storage decision is then preferably implemented by a storage manager <b>86.</b>

0050Preferably, storage manager <b>86</b> is able to select from a plurality of different storage options. More preferably, these different storage options include at least two different storage options having different types of accessibility. Examples of storage options having different types of accessibility include but are not limited to on-line storage, near-line storage and off-line storage. The type of storage media which is used for any particular storage option is not limited according to the present invention. For the purposes of illustration only and without any intention of being limiting, storage manager <b>86</b> is shown as being able to select from a plurality of storage devices, shown as an on-line storage device <b>88</b> and an off-line storage device <b>90.</b> Of course, other types of storage devices and/or systems could be used in place of, or in addition to, these examples of storage devices.

0051Rule engine <b>84</b> is optionally and preferably able to feed back information to format analyzer <b>62,</b> for improving the performance of format analyzer <b>62.</b>

0052A non-limiting example of the operation of system <b>50</b> may be performed as follows. System <b>50</b> could optionally be implemented at a service center which processes services requests from customers remotely, such that the customer is not physically present at the service center. The customer therefore contacts service center personnel, for example through voice communication (such as a telephone call for example).

0053A plurality of business rules has been defined and implemented by rule engine <b>84</b>, which could optionally include the following rules: a record is kept for every customer contact that refers to a financial transaction for at least three months even if no transaction occurred, and is kept for each contact in which a financial transaction occurred for at least seven years. In addition, the record for each contact resulting in an actual financial transaction is first stored in on-line storage for one month, and then in an off-line storage for the remainder of the term to seven years.

0054Once the customer has contacted a service center operator, for example through the telephone, data is provided through audio source <b>56</b> as an example of input source <b>52</b>. This captured audio data is analyzed, for example in order to determine if the financial transaction occurred during the contact. If such a transaction occurred, then metadata associated with the captured audio data indicates such an occurrence. Format analyzer <b>62,</b> and particularly audio analyzer <b>79,</b> preferably analyzes the captured audio data to obtain such metadata.

0055The data itself from the call is preferably handled according to one or more business rules, which may optionally be defined manually and/or generated automatically, through the operation of rule engine <b>84.</b> Preferably, rule engine <b>84</b> then generates an action to be performed by storage manager <b>86.</b> For example, storage manager <b>86</b> may optionally store the data from the call, migrate the data to a new type of storage, or delete the data, or any other action or any combination of actions which should be performed according to one or more events. Thus, rule engine <b>84</b> is able to generate one or more instructions for execution by storage manager <b>86.</b>

0056While the invention has been described with respect to a limited number of embodiments, it will be appreciated that many variations, modifications and other applications of the invention may be made, within the scope of the appended claims.