WO2017144953A1

WO2017144953A1 - System for providing contextually relevant data in response to interoperably analyzed structured and unstructured data from a plurality of heterogeneous data sources based on semantic modelling from a self-adaptive and recursive control function

Info

Publication number: WO2017144953A1
Application number: PCT/IB2016/051062
Authority: WO
Inventors: Claus KARTHE; Michael TANJANGCO; Marlo Marlito DOMINGO; Peter HEINCKIENS
Original assignee: Natural Intelligence Solutions Pte Ltd
Priority date: 2016-02-26
Filing date: 2016-02-26
Publication date: 2017-08-31

Abstract

The present utility model provides a computer-based system for providing contextually relevant output data based on semantic modelling derived from a self-adaptive and recursive control function. Structured and unstructured input data from heterogeneous data sources are processed and analyzed by an array of processing units. A set of rules is applied against the processed and analyzed data in order to generate measured process values from which contextually relevant output data may be generated. The contextually relevant output data is returned to the database system as part of the structured and unstructured input data for further processing and analysis by the recursive control function within a predetermined time period. A central processing unit and a graphics processing unit are configured to accelerate the data processing operations associated with the recursive control function.

Description

System for Providing Contextually Relevant Data in Response to Interoperably Analyzed Structured And Unstructured Data From a Plurality of Heterogeneous Data Sources Based on Semantic Modelling From a Self-Adaptive And Recursive Control Function

The present utility model generally relates to a computer-based system for providing contextually relevant output data. More particularly, it relates to a system for providing contextually relevant data in response to interoperably analyzed structured and unstructured data from a plurality of heterogeneous data sources based on semantic modelling from a self-adaptive and recursive control function.

Background of the Related Art

The traditional way of operating an enterprise comes with many burdens when it comes to data-driven decision-making. How data could be used effectively in the course of such decision-making has become dependent on "structured" data. From the conventional storage, processing, querying and up to retrieval operations, data are arbitrarily structured or "labelled" so that they can be effectively managed, and may serve as useful actionable intelligence for improving strategic and day-to-day business decisions.

Through the use of spreadsheets and relational databases, both of which are examples of applications commonly used for defining the structure of data such as data type (e.g., real, integer, Boolean) and data restrictions (e.g., string length, date format, null characters), data management and interchange have grown more efficient and streamlined for enterprises in the face of exponentially increasing data due largely to the structured data technology standards like SQL (Structured Query Language).

In the wake of "big" data however, structured data including those that are semi-structured appear to have been overshadowed by the emergence of totally unstructured data. The rapid increase in volume and speed of generation, not to mention the variation of data, both structured and unstructured, has resulted in information overload. Examining and selectively picking data from multiple disparate sources, understanding what they mean from various viewpoints or in various contextual environments, spotting trends and patterns now introduce challenges to businesses, especially in the way the same data is organized, analyzed, interpreted, and presented to decision-makers.

U.S. Patent Publication No. 20140171039 published on 19 June 2014 to Bjontegard Bernt Erik discloses an intelligent, self-learning and continually improving system which presents to a connected device offers, coupons, and information and then provides feedback from the connected device recording the effectiveness of the same offers, coupons and information, report this into a feedback loop and store this into a result database server which in turn reports this into a database that records this information and combines it with past recorded data and then reports this back into the historic and preference database providing a complete feedback loop with each interaction recorded and stored for repeated use when the same system is activated again thereby creating self-learning mechanism.

The above cited system further and specifically discloses a computer-implemented process for contextually intelligent mobile communication comprising: providing a mobile communication device; providing a plurality of sensors residing in said device; associating said device with a first user; providing first user past and historical data; generating contextually relevant output data for said first user with said device; gathering said contextually relevant output data on said device to create first user current contextually relevant output data; uploading said first user current contextually relevant output data via a wide area mobile communication network to a contextually intelligent server; matching said first user current contextually relevant output data with said first user past and historical data; generating feedback data for said first user that is contextually relevant to said user's current context and predictably useful to said user as said user enters a new context to form first user useful feedback data; transmitting to and displaying on said device said first user useful feedback data to provide said contextually intelligent mobile communication.

In essence, the above cited system is arguably characterized by a contextually intelligent, self-learning communication system since it is configured to combined unstructured data (e.g., social media status from Facebook and Twitter) and the structured data (e.g., data from the content and experience database and the historical data and demographics database) associated with the user, deliver content to the user in the form of recommendations, and measure actions taken by user on the content, and store the measured actions in the databases for further data analysis operations.

A problem associated with the cited prior system, however, is that the possibly indeterminate volume of both structured and unstructured data which undergo a closed-loop data processing cycle will certainly require high data processing speed. Increasing the physical quantities of data processing units associated with the prior system is one potential solution to said problem but the same necessitates higher power consumption and larger space requirement.

Thus, there remains an outstanding need to increase data processing speed for use in loop-based systems for providing contextually relevant output data based on possibly indefinite volume of structured and unstructured data from multiple sources across the globe.

Summary of Utility Model

The utility model provides a computer-based system for providing contextually relevant output data based on semantic modelling derived from a self-adaptive and recursive control function. The system includes a database system into and from which data objects can be stored and retrieved, respectively, from a plurality of heterogeneous data sources such as source files, electronic devices, and electronic platforms by an array of processing units through a communication network.

The processing units interoperably analyze the data objects which include structured input data from the source files and unstructured input data from the electronic devices and electronic platforms. The processing units have a memory system with a computer-executable control function which can be operative to pass the structured and unstructured input data to data processing operations.

The data processing operations include: (i) configuring a set of rules in relation to a desired reference value based on user input; (ii) processing and analyzing the structured and unstructured input data; (iii) applying the set of rules against the processed and analyzed structured and unstructured input data in the database system to generate measured process values; (iv) generating contextually relevant output data associated with the measured process values; (v) returning the contextually relevant output data to the database system as part of the structured and unstructured input data; and (vi) iterating steps (ii) to (v) above within a predetermined time period.

The processing units include a central processing unit and a graphics processing unit which are configured to accelerate the data processing operations associated with the control function. Any of the data processing operations which are memory-intensive can be loaded on a further memory system associated with the graphics processing unit. Any remaining data processing operations which are not memory-intensive may be retained in the memory system of the central processing unit.

The graphics processing unit of the array of processing units may include Nvidia^TM's GRID server, Nvidia^TM Tesla™ GPU, Nvidia^TM's series of GeForce^TM, and Nvidia^TM GPU Boost, to name a few. Through the use of these families of graphics processors, speed of data processing operations associated with big data analytics can be optimized without significantly increasing power consumption and without the need to deploy multiple processing units that could occupy large spaces.

The provision of allocating the memory-intensive data processing operations to the graphics processing unit increases data processing speed for use in loop-based delivery of contextually relevant output data, such as those characterized by steps (ii) to (v) above, which may involve processing of possibly indefinite volume of structured and unstructured input data from the plurality of heterogeneous data sources.

For a better understanding of the utility model and to show how the same may be performed, preferred implementations thereof will now be described, by way of non-limiting examples only, with reference to the accompanying drawings.

Fig.1

is a general block diagram of a computer-based system for providing contextually relevant output data of the present utility model.

Fig.2

is a flowchart of exemplary data processing operations of the system of Figure 1.

Fig.3

is an illustration of accelerating data processing operations of the system of Figure 1.

Fig.4

is block diagram illustrating interoperably analyzed structured and unstructured input data of the system of Figure 1.

Fig.5

is an exemplary computing environment into which the system of Figure 1 may be implemented.

Detailed Description of Preferred Implementations

Referring to Figure 1, there is shown a general block diagram illustrating a computer-based system for providing contextually relevant output data based on semantic modelling derived from a self-adaptive and recursive control function in accordance with one or more preferred implementations of the present utility model. The computer-based system is generally designated by reference numeral 100 throughout the ensuing description of preferred implementations or embodiments of, or best modes of carrying out, the present utility model.

The system 100 includes a database system 102 into and from which data objects can be stored and retrieved by an array of processing units 104. The database system 102 may include a range of database solutions including, but not limited to, MySQL (an open-source relational database management system), MongoDB (an open-source NoSQL database), PostgreSQL (an object-relational database management system (ORDBMS), CouchDB (an open-source database for the real-time web), RethinkDB (another open-source database for the real-time web), OrientDB (an open-source multi-model NoSQL database management system written in Java), and Redis (an in-memory database).

The array of processing units 104, which is used for storing in the database system 102 the data objects, is configured to interoperably analyze the same data objects originating from a plurality of heterogeneous data sources 106. Third-party electronic devices 108 and third-party electronic platforms 110 may constitute the heterogeneous data sources 106 which transmit the data objects to the array of processing units 104 over a communication network 112 in accordance with any suitable past and present communication protocol and/or communication standards set by professional organizations such as the IEEE (Institute of Electrical and Electronics Engineers).

The communication network 112 through which the data objects can be transmitted to the array of processing units 104 from the heterogeneous data sources 106, or vice versa, may be a wireless communication network, an external positioning method, or a Wireless Fidelity access point. The wireless communication network may include a CDMA network, a 3G network, a 4G/LTE network, and a 5G network.

Preferably, the communication network 112 is an IP (Internet Protocol) suite and its application layer protocols include the HTTPS (HyperText Transfer Protocol Secure) for secure communication of data to and from the array of processing units 104. It is to be understood and appreciated, however, that other types of network may be utilized in the process of communication between the array of processing units 104 and any one of the heterogeneous data sources 106. For example, the communication network 112 may alternatively be a wide area network (WAN), a metropolitan area network (MAN), a wireless local area network (WLAN), or a local area network (LAN).

The heterogeneous data sources 106 include source files 114 for providing structured input data in the data objects. The source files 114 carrying the structured input data that can be ingested into the database system 102 may be characterized by generic data sources such as file formats or data interchange formats like JSON, CDR, CSV, XLS, XML, HTML, SQL, RSS, and RDF. These file formats, which contain the structured input data, may be processed by the array of process unit units 104 in a well-defined manner to make them suitable for use in the system 100 for providing contextually relevant output data of the present utility model. These file formats may be provided by enterprises and businesses.

The heterogeneous data sources 106, as mentioned, include the electronic devices 108 and the electronic platforms 110 for providing unstructured input data in the data objects. The electronic devices 108 and electronic platforms 110 carrying the unstructured input data that can be ingested into the database system 102 may be include or otherwise characterized by software interfaces such as application programming interfaces and web services which are capable of providing access to network-based services or cloud-based services.

The electronic devices 108 can be selected from a group comprising of a desktop computer, a laptop, a mobile phone, a tablet, a phablet, and a beacon. Essentially, most of these devices are operable to run and manage web-based or mobile-based applications which are capable of capturing various data objects which may be in the form of either text data, image data, audio data, video data, or location data. Further, any of these devices may include peripheral devices such as cameras or image capturing and processing devices, GPS (Global Positioning System) devices, microphones or audio recording and processing devices, and interface devices such as touch screens, all of which may act as sensors suitable for capturing various forms of data.

The electronic platforms 110 can be selected from a group comprising of a social media platform (such as Facebook^TM, Twitter^TM, Google Plus^TM), an e-commerce platform, a gaming platform, content publishing platform, search engine platform, a digital marketing platform, e-mail delivery platform, VoIP (Voice over Internet Protocol) platform, a messaging platform, an multimedia streaming platform, and a multimedia hosting platform (Instagram^TM). Essentially, these platforms are likewise capable of capturing various data objects which may be in the form of either text data, image data, audio data, video data, or location data. Most of these platforms capture the various forms of data by way of directly receiving input data from a user such as comments, feedback, links websites, and tags.

The array of processing units 104 of the system 100 of the present utility model further includes a control function 116 in the form of computer-executable instructions which may reside in a memory system 118. The memory system 118 which may include one or more memory devices that are in communication with one another in the array of processing units 104. More particularly, the memory system 118 may include one or more of a volatile random access memory (RAM), a non-volatile read-only memory (ROM), a flash memory, and a ferroelectric RAM (F-RAM), all of which are not illustrated in the drawings as they are well known in the art of computing.

The memory system 118 is in operative communication with the array of processing units 104. The control function 116 is arranged for execution by any one or more processing units 104 of the array of processing units 104 from the memory system 118. When the control function 116 is executed by any of the processing units 104 included in the array of processing units 104, the control function 116 is operative to pass the structured and unstructured input data to data processing operations.

The data processing operations are separately exemplified in a flowchart in Figure 2 in accordance with one or more preferred implementations of the present utility model. The system 100 illustrated in Figure 1 is now taken in conjunction with the data processing operations exemplified in Figure 2. As shown in block 200, the data processing operations may commence with the step of configuring a set of rules in relation to a desired reference value based on a user input. This user input indicating the desired reference value may be any value that the user would like to serve as a basis in searching or mapping contextually relevant information, or meaningful information, through the structured and unstructured input data in the database system 102.

The set of rules that can be configured in relation to the desired reference value may be characterized by a logic which may be configured to execute an action only when a certain condition is satisfied. This set of rules may behave similar to a pattern matching algorithm for implementing production rule systems or Rete matching algorithm which may include condition and action statements, a fact, a session and flow. In this regard, an action may contain events such as bash execution, HTTP requests, audit logs, push notify and anything which could be considered as an outbound system event. Facts could be of simultaneous instances, and these create a support for processing data hypercubes where a data pattern may be discovered.

The data processing operations may then include the step of processing and analyzing the structured and unstructured input data as shown in block 202 and, subsequently, the step of applying the set of rules against the processed and analyzed structured and unstructured input data in the database system 102 to generate measured process values as shown in

blocks

204 and 206. The steps in the previous block 202 may require access to the database system 102 from which the structured and unstructured input data can be retrieved and then subjected to the set of rules. It is to be understood and appreciated that the database system 102 may include one or more SQL or noSQL databases.

The set of rules may correspond to business rules that can be composed, stored in the memory system 118, executed by the array of processing units 104 from the memory system 118, and reused for subsequent execution either within parametric loop bounds or reconstructed parametric loop bounds that can be expressed as a computational model for generating a semantic model based on selected conditions, parameters, and auxiliary variables.

The measured process values, which fundamentally originate from any or both of the structured input data of the source files 114 and unstructured input data from the electronic devices 108 and electronic platforms 110, are indicative of behavior and/or pattern of the structured and unstructured input data in respect of the desired reference value and in relation to the set of rules (e.g., business rules) applied against them. This behavior and/or pattern of the structured and unstructured input data are critical in many aspects such as, for example, semantic modelling and predicting events based on the semantic model.

The data processing operations may proceed with the step of generating contextually relevant output data associated with the measured process values as shown in block 208, and may then progress to the step of returning the contextually relevant output data to the database system 102 as part of the structured and unstructured input data, as shown in subsequent block 210. In particular, the contextually relevant output data generated in the previous block 208 may be returned to the part or parts of the database system 102 which stores unstructured input data from the electronic devices 108 and electronic platforms 110.

The data processing operations may further include the step of iterating steps (ii) to (v) above, or the equivalent steps in the

previous blocks

202, 204, 206, 208 and 210, within a predetermined time period, as now shown in block 212. At decision block 214, while the predetermined time period does not expire, a loop is formed by the aforesaid iterated steps. Otherwise, or if the predetermined time period expires, the data processing operations may further be arranged to generate a human-readable representation of the contextually relevant output data or, simply, a visual report, as shown in block 216.

The data processing operations may further comprise transmitting the human-readable representation of the contextually relevant output data from the array of processing units 104 to one or more data communication devices over the communication network 112. These data communication devices may be those that are operable by businesses or may correspond to the electronic devices 108. In addition, the human-readable representation of the contextually relevant output data may also be transmitted to any one of the electronic platforms 110, depending on predetermined configurations and/or on request from users connected to the electronic platforms 110.

One or more processing units of the array of processing units 104 include at least one central processing unit 120 and at least one graphics processing unit 122 configured to accelerate the data processing operations associated with the control function 116. Any of the data processing operations which are memory-intensive can be loaded on a further memory system 124 associated with the graphics processing unit 122. Any remaining data processing operations which are not memory-intensive, or which have low memory consumption, may be retained in the memory system 118 associated with the central processing unit 120.

Figure 3 shows an illustration of accelerating the data processing operations in accordance with one or more preferred implementations of system, as shown in Figure 1, of the present utility model. The graphics processing unit 122 of the array of processing units 104 may include Nvidia^TM's GRID server, Nvidia^TM Tesla™ GPU, Nvidia^TM's series of GeForce^TM, and Nvidia^TM GPU Boost, to name a few. Through the use of these families of graphics processors, speed of data processing operations associated with big data analytics can be optimized without significantly increasing power consumption and without the need to deploy multiple processing units that could occupy large spaces.

Through the provision of allocating the memory-intensive data processing operations to the graphics processing unit 122, the data processing speed associated with the data processing operations is increased, and this increase in the data processing speed is suitable for use in the loop-based delivery of contextually relevant output data, such as that characterized by steps in blocks 202 to 210 in Figure 2, which may involve processing of possibly indefinite volume of the structured and unstructured input data from the heterogeneous data sources 106.

The loop formed by the iteration step in the block 212, in effect, measures the actions taken on the contextually relevant output data which correspond to the measured process values returned to the database system 102, as new structured and/or unstructured input data corresponding to such action may be gathered from any of the heterogeneous data sources 106 (i.e., the source files 114, the electronic devices 108, and the electronic platforms 110) and can be combined with previously stored structured and unstructured input data in the database system 102.

Referring to Figure 4, there is shown a block diagram illustrating interoperably analyzed structured and unstructured input data in accordance with one or more preferred implementations of the present utility model. Preferably, the database system 102 includes two databases, a first database 400 which stores processed unstructured input data, and a second database 402 which stores parsed structured input data. The first database 400 may be a noSQL database while the second database 402 may be an SQL database.

The unstructured input data that can be stored in the first database 400 may undergo data processing operations including, but not limited to, location or proximity processing 404, image and video processing 406, audio processing 408, and person and product profiling 410. The structured input data that can be stored in the second database 402, on the other hand, may likewise undergo data processing operations including, but not limited to, real-time analytics 412, event parsing 414, and data parsing 416.

Referring now to Figure 5, there is an exemplary computing environment into which the system of Figure 1 may be implemented in accordance with one or more preferred implementations of the present utility model. The computing environment preferably includes the inbound data 500 which may correspond to the file format of the data that the system for providing contextually relevant output data of the present utility model can accept, read, and processed. As mentioned in Figure 1, these inbound data can be included and/or processed in the file format of HTTP, CDR, CSV and XML, among others.

The inbound files can be ingested into a high speed parsing engine 502 which may be a virtual processor loaded with various software components such as session control, SQL parser, process optimizer, and task dispatcher, among others. These software components operably cooperating with one another permit the system for providing contextually relevant output data of the present utility model to parse almost every kind of data that is known in the art to which the system of the present utility model belongs.

The parsing engine 502 of the computing environment of the system for providing contextually relevant output data of the present utility model is characterized by a complex event processing or method of tracking and analyzing (i.e., processing) streams of information (i.e., inbound data) about things that took place (i.e., events) and deriving a conclusion or inferred data from the analyzed streams of information. Stated differently, the parsing engine 502 is arranged to combine data from multiple sources in order to infer events or patterns that suggest more complicated circumstances, instances, and observations. The goal of complex event processing is to identify meaningful events and respond to them as quickly as possible.

The parsing engine 502 preferably includes a mediation layer 504 as a subsystem which handles receiving or retrieving, preparing, or parsing, the inbound data to be subjected to the complex event processing. The mediation layer 504 may include docks and transformation core functions. The docks are the adaptors for different sources of inbound data. These sources, by way of examples, may be characterized by any one of or a combination of any of TCP sockets, web sockets, SFTP, RPC, and web service.

The TCP (Transmission Control Protocol) sockets are the data sources that use pre-defined TCP stream protocols. The web sockets are the data sources that use predefined web sockets stream protocols. The SFTP (SSH or Secure Shell File Transfer Protocol) is the data source acquired via PUT or GET SFTP operations. The RPC (Remote Procedure Call) is the data source acquired through remote procedure call using pre-defined functions like CORBA that may be obtained by a service request from program hosted in a remote computer. The web service, finally, is the data source acquired via arbitrary web service.

The transformation, on one hand, is the function that transforms data sources into format of data that can be accepted, read, and processed by an events processor 506 of the high speed parsing engine 502. Instance's methods can be used in specific permutations depending on the source format. For example, "decrypt" can be used to transform process for encrypted source data. "Convert" can be used to transform process for non-plain text source data files with special encoding. "Tokenize" can be used to transform process for delimited plain text data files. "Offset parse" can be used to transform process for fixed width source data files.

The events processor 506 is the core of the complex event processing of the illustrated computing environment. The event processor 506, to reiterate, performs event processing (i.e., tracking and analyzing) the streams of information (i.e., structured input data) about things that happened or took place (i.e., events) and deriving a conclusion from the processed streams of information.

The parsing engine 502 of the computing environment also includes a real-time analytics engine 508 which is configured to have the capacity to use, all available enterprise data and resources when they are needed. The real-time analytics engine 508 consists of dynamic analysis and reporting, based on the inbound data into a system for providing contextually relevant output data of the present utility model less than one minute before the actual time of use.

Visualization tools 510-a, 510-c may also constitute the computing environment and may be grouped into reports 510-c and dashboards 510-a. The visualization tools 510-a, 510-c are arranged to deliver to an application layer database 512 based on a predetermined format for further processing or for display. The output format is "an easy to read, single page, real-time user interface, showing a graphical presentation of the current status (snapshot) and historical trends of an enterprise’s key performance indicators to enable instantaneous and informed decisions to be made at a glance.

The presentation layer database 512 is preferably an open source, BSD (Berkeley Software Distribution) licensed, advanced key-value cache and store. It is often referred to as a data structure server since keys can contain strings, hashes, lists, sets, sorted sets, bitmaps and hyperloglogs. This database stores all data for visualization. The presentation layer database 512 can be a multi-purpose database of various events and data, their meanings, and their patterns.

The other database 514, a high-performance database, may be considered as a genome vault. The unstructured input data which are processed and analyzed (i.e., by subjecting them to one or more computational models for semantic analysis) are preferably stored in the other database 514. The application of any one or more of the computational models against the unstructured data may be controlled by a deep-learning engine 516 which corresponds to the control function illustrated in Figure 1.

The computing environment further includes a high performance rules engine 518 which is configured to aggregate the processed and analyzed structured input data from the parsing engine 502 and as well as the processed and analyzed unstructured input data from the other database or genome vault 514. The rules engine 518 is operable to apply campaigns, rules and logic against the processed and analyzed structured and unstructured input data in order to generate the measured process values illustrated in Figure 1.

The rules engine 518 is preferably developed with NodeJS that executes one or more business rules in a runtime production environment. A task composer 520, which acts as the GUI (graphical user interface) of the rules engine rules engine 518, may also constitute the computing environment and could be a wrapper developed with PHP and MySQL. Through the task composer 520, rules engine 518 are visualized for easier usability.

The computing environment may also include a workflow engine 522 which executes APIs (application programming interfaces) from any of the third-party electronic platforms that are in communication with the array of processing units illustrated in Figure 1. In essence, the unstructured input data from various data sources, which may include sensors in operative communication with the third party electronic platforms, may flow through the workflow engine 522 in order for them to reach the rules engine 518 and cause generation of contextually relevant output data.

It is to be understood and appreciated that the execution of computer-executable instructions can be done on appropriate computing hardware and number of computing hardware, and that the illustrated core engines are merely segregated based on their intended functions for the sake of illustrating how they are relevant to the implementation of one or more preferred embodiments of the system for providing contextually relevant output data of the present utility model.

It should be readily appreciated by a person skilled in the art of computing that the illustrated engines can be fewer or greater in number, as it is well known in the art that program codes representing various functions of different engines or modules can be combined or segregated in any suitable but efficient manner insofar as program execution is concerned.

Referring to Figure 6, there is shown a detailed block diagram illustrating an exemplary computing system which can be used as a server apparatus 600 associated with the array of processing units illustrated in Figure 1. As illustrated, the server apparatus 600 preferably includes a central processing component 602, a graphics processing component 604, a memory component 606, a storage component 608, an input component 610, an output component 612, and a network interface component 614.

The exemplary computing system, which can be used as the server apparatus for use in herein disclosed system for providing contextually relevant output data of the present utility model, also includes a local host bus 616 and a local I/O (input/output) bus 618. A local bus controller 620 provides a bridge between the local host bus 616 and the local I/O bus 618. The local bus controller 620 generates the command to control the sequencing of the computer-executable instructions as fully described in Figure 1.

The illustrated computing system of the server apparatus 600 includes the central processing component 602 and a memory controller 622 which are in communication with one another through the local host bus 616. The graphics processing component 604 and the memory controller 622 are likewise in communication with one another through the local host but 616.

The memory component 606 which may comprise one or more memory devices is connected with the memory controller 622 and is where software, applications, or computer programs 624 associated with the system for providing contextually relevant output data of the present utility model may reside. Any of the central processing component 602 and the graphics processing component 604 may execute the computer-executable instructions or the computer programs 624 from the memory component 606 through the memory controller 622 and the local host bus 616.

The computer programs associated with the system for providing contextually relevant output data of the present utility model may be manipulated by a human user through the components that are connected with the local I/O bus 618 that is communication with the local host bus 616 through the local bus controller 620. Any changes made in the computer programs are recorded on the memory component 606 and are reflected on a display screen of a remote computing device through the network (or communication) interface component 614 connected with the local I/O bus 618.

While the present utility model has been described with respect to a limited number of implementations, those skilled in the art, having benefit of this disclosure, will appreciate that other implementations can be devised which do not depart from the scope of the utility model as disclosed herein.

Claims

A computer-based system for providing contextually relevant output data based on semantic modelling derived from a self-adaptive and recursive control function, the system comprising: a database system; an array of processing units for storing in the database system and interoperably analyzing data objects from a plurality of heterogeneous data sources through a communication network, the plurality of heterogeneous data sources including source files for providing structured input data in the data objects, the plurality of heterogeneous data sources further including at least electronic devices and electronic platforms for providing unstructured input data in the data objects; and a control function residing in a memory system of the array of processing units and which, when executed by any one or more processing units of the array of processing units from the memory system, is operative to pass the structured and unstructured input data to data processing operations comprising: (i) configuring a set of rules in relation to a desired reference value based on a user input; (ii) processing and analyzing the structured and unstructured input data; (iii) applying the set of rules against the processed and analyzed structured and unstructured input data in the database system to generate measured process values; (iv) generating contextually relevant output data associated with the measured process values; (v) returning the contextually relevant output data to the database system as part of the structured and unstructured input data; and (vi) iterating steps (ii) to (iv) above within a predetermined time period, wherein the one or more processing units of the array of processing units include at least one central processing unit and at least one graphics processing unit configured to accelerate the data processing operations associated with the control function.
The system according to claim 1, wherein the structured input data against which the set of rules is applied include inferred data.
The system according to claim 1, wherein the data processing operations further comprises generating a human-readable representation of the contextually relevant output data.
The system according to claim 3, wherein the data processing operations further comprises transmitting the human-readable representation of the contextually relevant output data to one or more data communication devices over the communication network.
The system according to claim 1, wherein the at least one graphics processing unit includes a further memory system.
The system according claim 5, wherein the data processing operations which are memory-intensive are loaded on the further memory system associated with the at least one graphics processing unit.
The system according to claim 1, wherein the source files are in data interchange format.
The system according to claim 1, wherein the electronic devices are selected from a group comprising of a desktop computer, a laptop, a mobile phone, a tablet, a phablet, and a beacon.
The system according to claim 1, wherein the electronic platforms are selected from a group comprising of a social media platform, an e-commerce platform, a gaming platform, content publishing platform, search engine platform, a digital marketing platform, e-mail delivery platform, VoIP (Voice over Internet Protocol) platform, a messaging platform, an multimedia streaming platform, and a multimedia hosting platform.
The system according to claim 1, wherein the unstructured input data include any one or more of text data, image data, audio data, video data, and location data.