Patent Application
20200111009
The disclosure herein generally relates to data analytics, and, more particularly, to automation of data analytics in an Internet of Things (IoT) platform.
By definition (advanced) analytics refers to theories, technologies, tools, and processes that enable an in-depth understanding and discovery of actionable insights in big data, which cannot be achieved by traditional data analysis and processing theories, technologies, tools, and processes. Major problem in current existing solutions involve inaccurate cataloging of information to handle problems of different domains, inability to fine tune hyper parameters and metric for processing given set of input sensor data. Further, information gathered from across domains, and subject matter experts involve lots of jargon which are difficult to interpret and build relationship models. This has severe impact over data analytics in an IoT platform as the incoming data is large in volume and probably in different formats which can further lead to inaccurate feature extraction and interpretation from input sensor data.
Embodiments of the present disclosure present technological improvements as solutions to one or more of the above-mentioned technical problems recognized by the inventors in conventional systems. For example, in one aspect, there is provided a processor implemented method for automatically performing data analytics in an IoT platform. The method comprises: receiving, input data comprising (i) type of problem and one or more solutions associated with the type of problem, and (ii) sensory information from a plurality of sensors deployed in an Internet of Things (IoT) platform, and (iii) domain knowledge from one or more sources, the domain knowledge is specific to the input data and corresponds to one or more domains, wherein the one or more solutions are indicative of a performance metric; representing, using an ontology based format technique, the input data as a plurality of graphical knowledge representations; determining, relationship between (i) the plurality of sensors and associated attributes and (ii) the domain knowledge across the plurality of graphical knowledge representations; applying, one or more Machine Learning (ML) models on the determined relationship; optimizing the one or more ML models based on the performance metric; and determining, using the optimized ML models, root cause analysis and interpretable root cause information associated with the root cause analysis, from the relationship determined across the plurality of graphical knowledge representations. In an embodiment, the interpretable root cause information varies based on the domain knowledge obtained from the one or more sources.
In an embodiment, the step of receiving, input data is preceded by constructing domain knowledge graph for each of the one or more domains and identifying relationship among information across the one or more domains.
In an embodiment, the step of determining, relationship between (i) the plurality of sensors and associated attributes and (ii) the domain knowledge across the plurality of graphical knowledge representations comprises identifying jargon in the domain knowledge across the plurality of graphical knowledge representations; and translating the jargon into interpretable data using one or more translation techniques. In an embodiment, the jargon may be translated into the interpretable data at one or more levels comprising a Lexicon level, a word level, a sentence level, a grammar level, or combinations thereof; and constructing a lexical database for each of the one or more domains, and wherein the constructed lexical database comprises a mapping between the interpretable data.
In another aspect, there is provided a system for automatically performing data analytics in an IoT platform. The system comprises a memory storing instructions; one or more communication interfaces; and one or more hardware processors coupled to the memory via the one or more communication interfaces, wherein the one or more hardware processors are configured by the instructions to: receive, input data comprising (i) type of problem and one or more solutions associated with the type of problem, and (ii) sensory information from a plurality of sensors deployed in an Internet of Things (IoT) platform, and (iii) domain knowledge from one or more sources, the domain knowledge is specific to the input data and corresponds to one or more domains; represent, using an ontology based format technique, the input data as a plurality of graphical knowledge representations; determine, relationship between (i) the plurality of sensors and associated attributes and (ii) the domain knowledge across the plurality of graphical knowledge representations; apply, one or more Machine Learning (ML) models on the determined relationship; optimize the one or more ML models based on the performance metric; and determine, using the optimized ML models, root cause analysis and interpretable root cause information associated with the root cause analysis, from the relationship determined across the plurality of graphical knowledge representations. In an embodiment, the interpretable root cause information may vary based on the domain knowledge obtained from the one or more sources.
In an embodiment, the one or more hardware processors are configured by the instructions to construct, prior to receiving the input data, domain knowledge graph for each of the one or more domains and identify relationship among information across the one or more domains.
In an embodiment, the one or more hardware processors are configured by the instructions to identify jargon in the domain knowledge across the plurality of graphical knowledge representations, and translate the jargon into interpretable data using one or more translation techniques.
In an embodiment, the jargon may be translated into the interpretable data at one or more levels comprising a Lexicon level, a word level, a sentence level, a grammar level, or combinations thereof. In an embodiment, the one or more hardware processors are configured by the instructions to construct a lexical database for each of the one or more domains. The constructed lexical database may comprise a mapping between the interpretable data.
In yet another aspect, there are provided one or more non-transitory machine readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors causes a method for automatically performing data analytics in an IoT platform. The method comprises: receiving, input data comprising (i) type of problem and one or more solutions associated with the type of problem, and (ii) sensory information from a plurality of sensors deployed in an Internet of Things (IoT) platform, and (iii) domain knowledge from one or more sources, the domain knowledge is specific to the input data and corresponds to one or more domains, wherein the one or more solutions are indicative of a performance metric; representing, using an ontology based format technique, the input data as a plurality of graphical knowledge representations; determining, relationship between (i) the plurality of sensors and associated attributes and (ii) the domain knowledge across the plurality of graphical knowledge representations; applying, one or more Machine Learning (ML) models on the determined relationship; optimizing the one or more ML models based on the performance metric; and determining, using the optimized ML models, root cause analysis and interpretable root cause information associated with the root cause analysis, from the relationship determined across the plurality of graphical knowledge representations. In an embodiment, the interpretable root cause information varies based on the domain knowledge obtained from the one or more sources.
In an embodiment, the step of receiving, input data is preceded by constructing domain knowledge graph for each of the one or more domains and identifying relationship among information across the one or more domains.
In an embodiment, the step of determining, relationship between (i) the plurality of sensors and associated attributes and (ii) the domain knowledge across the plurality of graphical knowledge representations comprises identifying jargon in the domain knowledge across the plurality of graphical knowledge representations; and translating the jargon into interpretable data using one or more translation techniques. In an embodiment, the jargon may be translated into the interpretable data at one or more levels comprising a Lexicon level, a word level, a sentence level, a grammar level, or combinations thereof. In an embodiment, a lexical database for each of the one or more domains may be constructed, and wherein the constructed lexical database comprises a mapping between the interpretable data.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.
The accompanying drawings, which are incorporated in and constitute a part of this disclosure, illustrate exemplary embodiments and, together with the description, serve to explain the disclosed principles.
Exemplary embodiments are described with reference to the accompanying drawings. In the figures, the left-most digit(s) of a reference number identifies the figure in which the reference number first appears. Wherever convenient, the same reference numbers are used throughout the drawings to refer to the same or like parts. While examples and features of disclosed principles are described herein, modifications, adaptations, and other implementations are possible without departing from the spirit and scope of the disclosed embodiments. It is intended that the following detailed description be considered as exemplary only, with the true scope and spirit being indicated by the following claims.
As mentioned above (advanced) analytics refers to theories, technologies, tools, and processes that enable an in-depth understanding and discovery of actionable insights in big data, which cannot be achieved by traditional data analysis and processing theories, technologies, tools, and processes. Major problem in current existing solutions involve inaccurate cataloging of information to handle problems of different domains (e.g., media (social media), manufacturing, healthcare, and the like), inability to fine tune hyper parameters and metric for processing given set of input sensor data, etc. Further different sets of hyper parameters and metrics required to be tuned on a specific project requirement (e.g., like Remaining Useful Life prediction for a machine part or a human organ). This also involves lots of jargon in the manner information is received in large volume for data analytics. The existing solutions may also fail to apply machine learning when number of attributes and sensors that gives input are too many which make is difficult to identify whether there is any inherent structure lying with multiple attributes of multiple sensor or in other words whether there exists any mapping between the attribute of one sensor and other attribute of another sensor or not.
Embodiments of the present disclosure provide systems and methods for automatically performing data analytics in an IoT platform. The above technical problem is addressed by the present disclosure by providing a technical solution that involves defining a proper cataloguing service for analytics that support multiple views and exposing basic modules in multi layers of abstraction where each layer can access the contents of lower level by setting parameters of stitching multiple blocks under the bottom layer. These layers of abstractions are named as: Domain, Project, Tasks, Modules, and the like. Further knowledge of multiple stake holders is captured using ontology, represented under a graph for example, but is not limited to, Directed Acyclic Graph (DAG) and persisted in an offline mode for regeneration and reconstruct the DAG for further processing as and when required. Furthermore, all resultant features are obtained using machine learning technique(s)/models for example Rain Forest, Support Vector Regression, or deep learning technique(s)/models for example, but are not limited to Convolutional Neural Network, Long-Short Term Memory, Stacked Auto encoder decoder, and the like), wherein these features are interpreted in terms of business domain jargon. Moreover, one or more supervised, unsupervised, semi-supervised machine learning technique(s) or combinations thereof may be applied on the data being processed to identify the similarity among sensors and further applying tensor factorization on top of it to analyze the data. In an embodiment, the domain specific jargon may be represented in an ontology format, to identify the lexicons of a domain, further define a word-net for that domain and use that structure to map it to the jargon of other domain, which is very much required in a multi stakeholder scenario such as IoT analytics.
In a nutshell, the present disclosure describes interpreting the knowledge of a domain expert using some ontology or as a directed acyclic graph (DAG). The present disclosure further describes method of extracting the inherent structure within the data captured using multiple sensors using data analytics methodologies. Further, the present disclosure enables augmentation of domain knowledge constructed by correcting the expert knowledge based on information obtained by data driven approach as well as correcting features and dependencies obtained using analytical methods by applying expert knowledge obtained from individual or from other sources (e.g., websites, books, literature, and the like). Moreover, the present disclosure enables translation problem statement, root cause, from one knowledge to another using machine aided translation methodology/methodologies. For instance, domain knowledge of healthcare can be aimed to be translated into signal processing and machine learning domain.
Referring now to the drawings, and more particularly to
The I/O interface device(s) 106 can include a variety of software and hardware interfaces, for example, a web interface, a graphical user interface, and the like and can facilitate multiple communications within a wide variety of networks N/W and protocol types, including wired networks, for example, LAN, cable, etc., and wireless networks, such as WLAN, cellular, or satellite. In an embodiment, the I/O interface device(s) can include one or more ports for connecting a number of devices to one another or to another server.
The memory 102 may include any computer-readable medium known in the art including, for example, volatile memory, such as static random access memory (SRAM) and dynamic random access memory (DRAM), and/or non-volatile memory, such as read only memory (ROM), erasable programmable ROM, flash memories, hard disks, optical disks, and magnetic tapes. In an embodiment a database 108 can be stored in the memory 102, wherein the database 108 may comprise, but are not limited to information sensor data (pertaining to sensors deployed in the IoT platform/environment), domain knowledge from one or more sources, the domain knowledge being specific to the sensor input data and the like. More specifically, information pertaining to various sensors, data obtained from various subject matter experts, and sensor attributes, are stored in the database 108. In an embodiment, the memory 102 may store one or more technique(s) (e.g., machine learning model(s), and the like) which when executed by the one or more hardware processors 104 perform the methodology described herein. The memory 102 may further comprise information pertaining to input(s)/output(s) of each step performed by the systems and methods of the present disclosure.
Further say Domain 1 refers to a coder (another stakeholder). The coder due to his/her ability/experience and prior knowledge provides information on accelerometer, for instance, acceleration is double derivative of position with respect to Time. The coder also provides domain knowledge such as ‘velocity can be obtained by integrating the acceleration over time’. Other domain knowledge that the coder may further provide is that ‘angular velocity and angular momentum can be derived from radius and linear velocity’.
Using the above domain knowledge obtained from stakeholder 1 and stakeholder 2, cross domain knowledge may be constructed by the system 100. For instance, each of the bearing cage, motor and shaft has a particular working frequency. Here the abnormality, measured in terms of bearing cage, motor or shaft (obtained from domain 1—or from stakeholder 1) can be mapped and measured by obtaining angular velocity (defined in domain 2 by stakeholder 2). If the sensor attributes are orthogonal (in domain 1, say D1), their values are independent of each other (in domain 2, say D2). If task at hand is to choose a state among one or more states from a finite number of states, then it is a classification problem. Root cause can be identified from the most contributing feature in a classifier or by applying Bayesian Network. Different performance metrics for the task of classifications are Recall, Precision, Accuracy, F-Score, and the like can be measured.
Referring back to steps, in an embodiment of the present disclosure, at step 202, the one or more hardware processors 104 receive, input data comprising (i) type of problem and one or more solutions associated with the type of problem, and (ii) sensory information from a plurality of sensors deployed in an Internet of Things (IoT) platform, and (iii) domain knowledge from one or more sources, the domain knowledge is specific to the input data and corresponds to one or more domains. In an embodiment of the present disclosure, the one or more solutions are indicative of a performance metric (e.g., recall, precision, Fscore, and the like). The sources can be a website, books, literature, and the like, in one example embodiment. In an embodiment, considering the above domain knowledge scenarios, input data may comprise (i) type of problem may be defined as “to determine whether a bearing box is working properly or not and if not why it is not working in terms of accuracy”, (ii) data related to bearing box can be obtained from the accelerometer sensor which has three attributes namely acceleration in the direction of x, y, and z coordinates, and (iii) domain knowledge (e.g., say Domain 2 wherein Coder is the stakeholder) may be obtained in addition obtaining angular velocity from the frequency, wherein frequency can be obtained from Fast Fourier Transformation. Since the coder may not be able to anticipate all possible implementation knowledge, he may see the problem to solve. Once this knowledge is gathered, the initial knowledge-base is augmented with this new knowledge. In other words, the domain knowledge which was constructed initially prior to performing step 2, that domain knowledge gets augmented with this new knowledge.
In an embodiment of the present disclosure, at step 204, the one or more hardware processors 104 represent, using an ontology based format technique, the input data as a plurality of graphical knowledge representations. In the present disclosure, example of the plurality of graphical knowledge representations, may comprise but are not limited to, Directed Acyclic Graph (DAG). For instance, the input data is annotated as per the knowledge graph defined by the stakeholder (say stakeholder 1 which in this case is a mechanical engineer). Specifically “accelerometer is a type of sensor”, whose attributes are x, y, and z respectively. Now using the cross domain knowledge the problem is translated to apply Short Term Fourier Transformation (STFT) or Fast Fourier Transform (FFT) as a feature on the attributes of the sensor(s). In an embodiment, Short Term Fourier transformation (STFT) is a Fourier Transformation over a short time window. Hence the expressions STFT and FFT may be interchangeably used herein. In the present disclosure, STFT is applicable to the scenarios when FFT is applied over a short time of window, in one example embodiment.
In an embodiment of the present disclosure, at step 206, the one or more hardware processors 104 determine, relationship between (i) the plurality of sensors and associated attributes and (ii) the domain knowledge across the plurality of graphical knowledge representations. For instance, the relationship between the plurality of sensors and associated attributes may comprise, but is not limited to, Domain 1 defined that the sensor attributes are orthogonal.' Another instance wherein relationship between the domain knowledge across the plurality of graphical knowledge representations may comprise, but is not limited to, ‘as Domain 1 defined that the sensor attributes are orthogonal, the FFT should consider each axis as a separate feature’. In an embodiment, the step of determining, relationship between (i) the plurality of sensors and associated attributes and (ii) the domain knowledge across the plurality of graphical knowledge representations comprises identifying jargon in the domain knowledge across the plurality of graphical knowledge representations; and translating the jargon into interpretable data using one or more translation techniques (e.g., look-up of the cross-domain knowledge using techniques such as string similarity measures, or embedding-based techniques (or translation technique(s) such as word2vec or node2vec techniques). In an embodiment, wherein the jargon is translated into the interpretable data at one or more levels comprising a Lexicon level, a word level, a sentence level, a grammar level, or combinations thereof; and constructing a lexical database (e.g., a WordNet database) for each of the one or more domains, and wherein the constructed database comprises a mapping between the interpretable data. The lexical database may be stored in the memory 102, in one example embodiment. Jargon translation is explained by way of following example, wherein the example shall not be construed as limited the scope of the present disclosure. For instance, say, information from Mechanical Engineer is received as: Condition of a bearing box can be measured using accelerometer as a sensor. Each works under normal operating range under normal operating condition. Cross-domain translation can be achieved by: computing Velocity by taking integration of acceleration over time, and deriving angular velocity from linear velocity. Further frequency is derived (or obtained from the angular velocity, wherein short-time Fourier transform (STFT) is performed for analysis of frequency. Coder on the other hand says need to identify STFT over operating range for problem. The above example of jargon translation is therefore an output that can be referred as interpretable data.
In an embodiment of the present disclosure, at step 208, the one or more hardware processors 104 apply, one or more Machine Learning (ML) models on the determined relationship. Applying the one or more Machine Learning (ML) models on the determined relationship comprises executing, via the one or more hardware processors 104 the ML models on the determined relationship. For instance, cross domain knowledge gives an insight that the given problem statement is a classification and a root cause analysis problem. In an embodiment, the one or more ML models comprise, but are not limited to Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Support Vector Machine (SVM), Rain Forest (RF), and the like which may be applied depending on the ML task type (e.g., regression, anomaly detection, De-noising signal(s) having an inherent structure in it, and the like).
In an embodiment of the present disclosure, at step 210, the one or more hardware processors 104 optimize the one or more ML models based on the performance metric. The present disclosure, providing one or more users, the flexibility of defining a performance metric for different ML tasks depending on logic provided by domain expert(s)/subject matter experts (also referred as source). For each of the ML task, one or more feature reduction techniques may be used with a focus on optimizing the performance metric as described above. Few examples of feature reduction techniques being used are listed below, and shall not be construed as limited the scope of the present disclosure:
In an embodiment of the present disclosure, at step 210, the one or more hardware processors 104 determine, using the optimized ML models, root cause analysis and interpretable root cause information associated with the root cause analysis, from the relationship determined across the plurality of graphical knowledge representations. For instance, result of a classifier shows that the 33 Hertz is the most discriminating feature whereas normal operating frequency of shaft is 30-40 Hertz. So it can be concluded that the machine/bearing box is not working properly because of error in shaft. In the above example, the interpretable root cause information herein refers to ‘machine/bearing box is not working properly because of error in shaft’ and root cause analysis can be referred as ‘33 Hertz is the most discriminating feature whereas normal operating frequency of shaft is 30-40 Hertz which has resulted in machine not working properly’. Based on the domain knowledge obtained from the one or more sources (e.g., stakeholder 1 and stakeholder 2 from domains 1 and 2 respectively, the above interpretable root cause information was determined. It is to be understood by a person having ordinary skill in the art and person skilled in the art that interpretable root cause information that is being determined for a given problem (or type of problem or nature of problem under consideration) may vary based on the domain knowledge obtained from the one or more sources. For instance, the interpretable root cause information may not be same as above if the domain knowledge is obtained from say a stakeholder 3 (e.g., who is less qualified but has hands on experience in repairing bearing box and machine operating tools when compared with mechanical engineer). The steps 202 till 212 are automated by the present disclosure to thus enabling automation of data analytics in IoT platform as described herein by way of non-limiting examples.
The written description describes the subject matter herein to enable any person skilled in the art to make and use the embodiments. The scope of the subject matter embodiments is defined by the claims and may include other modifications that occur to those skilled in the art. Such other modifications are intended to be within the scope of the claims if they have similar elements that do not differ from the literal language of the claims or if they include equivalent elements with insubstantial differences from the literal language of the claims.
It is to be understood that the scope of the protection is extended to such a program and in addition to a computer-readable means having a message therein; such computer-readable storage means contain program-code means for implementation of one or more steps of the method, when the program runs on a server or mobile device or any suitable programmable device. The hardware device can be any kind of device which can be programmed including e.g. any kind of computer like a server or a personal computer, or the like, or any combination thereof. The device may also include means which could be e.g. hardware means like e.g. an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a combination of hardware and software means, e.g. an ASIC and an FPGA, or at least one microprocessor and at least one memory with software modules located therein. Thus, the means can include both hardware means and software means. The method embodiments described herein could be implemented in hardware and software. The device may also include software means. Alternatively, the embodiments may be implemented on different hardware devices, e.g. using a plurality of CPUs.
The embodiments herein can comprise hardware and software elements. The embodiments that are implemented in software include but are not limited to, firmware, resident software, microcode, etc. The functions performed by various modules described herein may be implemented in other modules or combinations of other modules. For the purposes of this description, a computer-usable or computer readable medium can be any apparatus that can comprise, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
The illustrated steps are set out to explain the exemplary embodiments shown, and it should be anticipated that ongoing technological development will change the manner in which particular functions are performed. These examples are presented herein for purposes of illustration, and not limitation. Further, the boundaries of the functional building blocks have been arbitrarily defined herein for the convenience of the description. Alternative boundaries can be defined so long as the specified functions and relationships thereof are appropriately performed. Alternatives (including equivalents, extensions, variations, deviations, etc., of those described herein) will be apparent to persons skilled in the relevant art(s) based on the teachings contained herein. Such alternatives fall within the scope and spirit of the disclosed embodiments. Also, the words “comprising,” “having,” “containing,” and “including,” and other similar forms are intended to be equivalent in meaning and be open ended in that an item or items following any one of these words is not meant to be an exhaustive listing of such item or items, or meant to be limited to only the listed item or items. It must also be noted that as used herein and in the appended claims, the singular forms “a,” “an,” and “the” include plural references unless the context clearly dictates otherwise.
Furthermore, one or more computer-readable storage media may be utilized in implementing embodiments consistent with the present disclosure. A computer-readable storage medium refers to any type of physical memory on which information or data readable by a processor may be stored. Thus, a computer-readable storage medium may store instructions for execution by one or more processors, including instructions for causing the processor(s) to perform steps or stages consistent with the embodiments described herein. The term “computer-readable medium” should be understood to include tangible items and exclude carrier waves and transient signals, i.e., be non-transitory. Examples include random access memory (RAM), read-only memory (ROM), volatile memory, nonvolatile memory, hard drives, CD ROMs, DVDs, flash drives, disks, and any other known physical storage media.
It is intended that the disclosure and examples be considered as exemplary only, with a true scope and spirit of disclosed embodiments being indicated by the following claims.
| Number | Date | Country | Kind |
|---|---|---|---|
| 201821037802 | Oct 2018 | IN | national |
This U.S. patent application claims priority under 35 U.S.C. § 119 to: India Application No. 201821037802, filed on Oct. 5, 2018. The entire contents of the aforementioned application are incorporated herein by reference.
