Patent Application
20250028715
Embodiments of the present disclosure relate to computing systems, and more specifically, to methods and systems for data standardization and comprehensive processing of queries.
Businesses, especially small-size businesses and medium-size businesses, and non-profit organizations often do not have sufficient computing resources and human personnel to develop comprehensive support services (e.g., software support, policies and procedures, tracking and reporting systems, etc.) fully in-house and have to rely on specialized outside developers and providers of these services. Such providers may furnish, to client businesses and organizations, various hardware and software computing resources (e.g., cloud-based and/or local) as well as subject-matter expertise that reduce administrative overhead and automate a significant number of support tasks, such as maintaining inventory, tracking deliveries and shipments, complying with rules and regulations, implementing data security and personnel/environment safety measures, managing human resources, and/or the like. Human resource management may include maintaining job descriptions, managing employee salaries and benefits, tracking job-related activities and relations of employees, supporting employee recruiting, monitoring employee satisfaction, keeping abreast of changes in the relevant industries and geographic areas, and/or the like.
The disclosure is illustrated by way of example, and not by way of limitation, and can be more fully understood with references to the following detailed description when considered in connection with the figures, in which:
Efficient provisioning of support services depends on the ability of a provider of such services to process information related to multiple clients (e.g., businesses) efficiently and comprehensively. Such information is often fragmented by industry, geography, state of client's operations, and so on. For example, even clients operating in the same industry and/or state may use very different terminology to refer to the same jobs, entities, operations, and procedures. For example, the same or similar jobs may be designated as “manager,” “supervisor,” “senior staff member,” “dispatcher,” “shift coordinator,” and/or the like. Work beyond regular business hours may be referred to as “overtime,” “double time,” “extra time,” and/or the like. Such data fragmentation may hinder the provider's ability to serve various clients. In particular, a client may be interested in learning about best business practices of human resource management, salary and benefits ranges of various jobs, on-site safety, and/or the like, but a search query that does not capture nuances and varieties of terminology being used may cause the provider to miss a significant slice of data, even if the data is available in the provider's data stores. Searching for and manually indexing various terms is a very expensive, slow, and inefficient process given large volumes of available information and the fact that such information may be constantly changing/updating as a result of new terminology being introduced, new clients served, and/or the like.
Aspects and implementations of the instant disclosure address the above-mentioned and other challenges of the existing technology by providing for systems and techniques capable of using machine learning for automated data standardization and efficient data query processing. In some embodiments, data entities collected from a variety of sources (e.g., clients, public sources, and/or the like) may be tokenized (split into segments having distinct semantic meanings), represented in the form of digital embeddings, and processed by one or more machine learning models (MLMs) to perform clustering of various units of data among clusters associated with specific semantic meanings. Clustering may be facilitated by associating anchor tokens (also referred to as, simply, anchors herein) with various clusters. Anchors represent some of the (actual or expected) data entities (units) having semantic values that are similar to the cluster's semantic value(s). In some embodiments, the MLMs that perform clustering may include a statistical model (SM) generating probabilities that a particular token is associated with various clusters. The SM can be trained using anchors. For example, individual clusters may be initiated (seeded) with one or more anchors. Various data units may then be processed by the SM and individual tokens in the data units may acquire, for each cluster, a score (weight) that is determined by a frequency with which the token is encountered in data units in combination with anchors of that cluster. The SM may be used together with one or more additional models. In some embodiments, a lexical model (LM), e.g., a model that deploys techniques of natural language processing model, may be used. The LM may be initially pre-trained using a general vocabulary and natural language training techniques. Subsequently, the LM may be further trained using data units labeled by the trained SM as ground truth. During classification of a new data unit, the data unit may be tokenized, digitized and processed by the SM and/or the LM, e.g., in parallel. Each of the models may generate separate predictions, e.g., probabilities, that the data unit or its individual tokens belong to specific clusters. The probabilities generated by individual models may be aggregated (e.g., averaged or otherwise weighted) and a cluster identified with a maximum probability may indicate a semantic value of the data unit (or its individual tokens). As a result, data units that include seemingly distinct but semantically close meanings (e.g., job titles, job descriptions, etc.) are associated with the same cluster. This facilitates that various database queries (e.g., searches and/or requests for specific categories of data) return comprehensive results that include multiple relevant items of information even when such items are expressed via different words (which in fact have the same or similar meanings).
The advantages of the disclosed techniques include but are not limited to quick and comprehensive queries for data, documents, content, structured and unstructured information that is available in digitized data stores. Correspondingly, efficient identification and retrieval of relevant information facilitates supplying clients with complete information for informed actions by the clients. The disclosed techniques may be used in workforce management (WFM) and human capital management (HCM) application and, in particular, including standardization of WFM/HCM for small and medium business entities, efficient identification and retrieval of jobs data, paycodes/counter data, transferring WFM/HCM practices developed by some organizations to other organizations in the same industry or to other industries, and/or the like. Furthermore, the disclosed techniques may be used for efficient cross-tenant management of tenants that use different business practices, terminology, data management techniques, and/or the like.
In some embodiments, DPs 120-j may store any suitable raw and/or processed data units that may be collected from any number of clients, e.g., businesses, public and private foundations, government agencies, non-profit organizations, institutions, associations, charities, partnerships, and/or the like. Different DPs 120-j may be serving (e.g., collecting information from) different geographic areas, states, industries, businesses of different types and sizes, and/or the like. In some embodiments, DPs 120-j may store information that includes job titles, job descriptions, salaries, benefits, employment policies, laws and government regulations, listing of services provided by clients to customers, inventories of goods, and/or any other suitable public and/or private data. DPs 120-j may store information together with various structures that tag, organize, and index the data. Data store 110 may store various data and metadata used and/or generated to facilitate processing of information collected in DPs 120-j. In some embodiments, data store 110 may store token clusters 111 (also referred to as simply clusters throughout this disclosure) that group various data units collected in DPs 120-j by semantic meaning. Each token cluster 111 may correspond to a specific semantic entity, e.g., “overtime pay,” but may be expressed using different words a multitude of different ways. Token clusters 111 may include references (e.g., pointers) to specific data units stored in one or more DPs 120-j that have been previously identified as belonging to various clusters. Token clusters 111 may be defined using anchors 112 (also referred to as anchor tokens herein). Anchors 112 may be example tokens that are commonly encountered (or expected to be encountered) in data units associated with semantic meaning of the respective clusters 111. In some embodiments, anchors 112 may be initially identified by system developers. Data store 110 may store a vocabulary 113, e.g., a corpus of words in a principal language used by at least some of the clients. Data store 110 may also store one or more foreign language (FL) vocabularies 114 with translations of at least some words of vocabulary 113. Data store 110 may also store a list of abbreviations 115 in the principal language and/or foreign language(s) and a list of acronyms 116 in the principal language and/or foreign language(s).
Any or some of data store 110 and/or DPs 120-j may be implemented in a persistent storage capable of storing files as well as data structures to perform identification of data, in accordance with implementations of the present disclosure. Data store 110 and/or DPs 120-j may be hosted by one or more storage devices, such as main memory, magnetic or optical storage disks, tapes, or hard drives, network-attached storage (NAS), storage area network (SAN), and so forth. Although depicted as separate from the server machine 130, data store 110 and/or DPs 120-j may be part of server machine 130, and/or other devices. In some embodiments, data store 110 and/or DPs 120-j may be implemented on a network-attached file server, while in other implementations data store 110 and/or DPs 120-j may be implemented on some other types of persistent storage, such as an object-oriented database, a relational database, and so forth, that may be hosted by a server machine 130 or one or more different machines coupled to server machine 130 via network 150.
In some embodiments, any of server machine 130 and/or client machine(s) 140 may include a desktop computer, a laptop computer, a smartphone, a tablet computer, a server, a scanner, or any suitable computing device capable of performing the techniques described herein. In some implementations, any of server machine 130 and/or client machine(s) 140 may be (and/or include) one or more computer systems 600 of
QSE 132 and/or QPE 138 may include (or may have access to) instructions stored on one or more tangible, machine-readable storage media of server machine 130 and executable by one or more processing devices of server machine 130. In one implementation, QSE 132 and/or QPE 138 may be implemented as a single component. In some implementations, QSE 132 and/or QPE 138 may each be a client-based application or may be a combination of a client component and a server component. In some implementations, QSE 132 and/or QPE 138 may be executed entirely on the client machine(s) 140. Alternatively, some portion of QSE 132 and/or QPE 138 may be executed on a client computing device while another portion of QSE 132 and/or QPE 138 may be executed on a server machine 130.
Tokens 204 of data unit 202 input into SM 134 may be processed by a token-anchor matching module 220. Token-anchor matching module 220 may compare tokens 204 with anchors 112 of various clusters 111. Each token 204 in data unit 202 that matches an anchor 112 of a particular cluster 111, e.g., a first cluster, may be assigned score 1 (or any other preferred value). The number n1 of tokens 204 that match anchors 112 of the first cluster may be divided by the total number n of tokens 204 in data unit 202 and the resulting score, S1=n1/n, may be assigned to tokens 204 that do not match any of the anchors 112 of the first cluster. Similar matching may be performed for other clusters 111, e.g., second cluster, third cluster, etc., and the corresponding scores S2, S3, etc., may be assigned to tokens 204 that do not match the anchors of the corresponding cluster, but are present in data unit 202 together with some of those anchors. As a result, each token 204 may be assigned a respective cluster score S1, S2, S3 . . . that indicates the affinity of the corresponding token to the first cluster, second cluster, third cluster, and so on.
As additional data units 202 are processed by SM 134, a set of cluster scores {Sj} of various tokens may be further modified in view of detected use of the tokens, which may be determined by a token use statistic module 230. More specifically, the second time a given token T is encountered in a new data unit, token T may already have a score Sj that characterizes affinity of token T to jth cluster. The average of jth cluster scores Sj
unit for all tokens of the new data unit may be computed and used to update the score Sj of token T. In one non-limiting embodiment, the updated score Sj-updated may be computed as follows:
where N is the number of data units with token T that have been encountered previously. As a result, the more frequently token T is encountered in the same data unit 202 with one of anchors 112 of jth cluster, the higher the score will be. In particular, if token T is always encountered with one of these anchors, score Sj will approach 1 after a large number of such encounters. Conversely, if token T is rarely encountered in the same data unit 202 with one of anchors 112 of jth cluster (or with other tokens that are frequently encountered with such anchors), the corresponding score Sj will tend to 0 (or a small value)
The above example is intended to be illustrative and not limiting. In other embodiments, various other schemes of assigning initial scores to tokens and updating scores of tokens may be used. For example, scores Sj may be limited by a certain maximum value Smax (e.g., Smax=0.5, 0.6, and/or the like). After token score computation module 240 has computed scores of tokens 204 encountered for the first time and/or updated scores of tokens previously encountered (e.g., based on information provided by token use statistics module 230), tokens and token scores may be stored in a token store 210 of data store 110. The stored information for token T may include token T itself, the number of data units in which token T has been encountered, the number of times token T has been encountered together with at least one of the anchors of each cluster, and/or the like. In some embodiments, token store 210 may store additional data related to correlations of use of different tokens, e.g., a number of times token T has been used together with various other tokens T′, T″, etc., that are not anchor tokens.
Query 302 may include a data unit 304 and any additional associated information 306. In some embodiments, data unit 304 may include information that is processed by machine learning models as part of workflow 300 whereas associated information 306 may be any additional information that is not processed by the models but may be used (e.g., by QPE 138) after the machine learning models have completed processing (e.g., classification) of data unit 304. For example, a data unit 304 may include a title of a document (e.g., “Job descriptions of Company A”) and associated information 306 may include job titles, job descriptions, and/or other information). In another example, associated information 306 may be referenced in data unit 304. For example, data unit 304 may include a request to collect information about email security policy adopted by public universities in a given state and compare the policies with the current policy of the client attached as associated information 306. In some instances, query 302 may include data unit 304 and does not include any associated information 306. For example, data unit 304 may include a request to “find a range of salaries of radiologists in rural areas of New England” with no associated information 306 provided.
Data unit 304 may undergo tokenization 310 that segments data unit 304 into one or more tokens 320. Tokens 320 may refer to any portion of data unit 304 having individual semantic meaning. In the last example, tokens 320 may include four tokens: “salaries” (or “range of salaries”), “radiologists,” “rural areas,” and “New England.” Tokenization 310 may remove filler words, prepositions, conjunctions, punctuation, and/or the like. Tokenization 310 may further include normalization, e.g., reducing plural forms of nouns to singular forms, bringing verbs to a preferred form (e.g., infinitive form), and/or the like.
Tokenization 310 may include foreign language (FL) token processing 314, e.g., translation of tokens from a foreign language (e.g., using FL vocabulary 114) to the primary language of data unit 304. In some embodiments, a web-based translator may be used for FL token processing 314 (e.g., via an appropriate Application Programming Interface available to QSE 132).
Tokenization 310 may further include abbreviation processing 315, which may replace various abbreviations (e.g., “Ovt,” “Mgr,” etc.) with regular words (e.g., “Overtime,” “Manager,” etc.), e.g., using a list of abbreviations 115. Tokenization 310 may also include acronym processing, e.g., replace various acronyms with expanded strings of words (e.g., replace “RN” with “registered nurse”), e.g., using a list of acronyms 116.
Tokenization 310 may also correct irregularly typed/formatted words. More specifically, tokenization 310 may include fused token correction 317, which may be based on vocabulary 113. For example, a string of tokens typed without a space, e.g., “stateagency,” may be changed to the correct string “state agency” and then split into two individual tokens. Tokenization 310 may further perform mistyped token correction 318, e.g., using vocabulary 113. For example, a mistyped token “Utube” may be changed to the correct token “YouTube.”
Tokens 320 may be processed by one or more models, e.g., SM 134 and LM 136. SM 134 may be trained as disclosed above in conjunction with
V
SM=(AS1, AS2, . . . ASM),
where M is the total number of classes, and subscript SM indicates that the embedding is generated by SM 134. Each of the M components ASj of the embedding VSM is representative of the likelihood of the data unit 304 belonging to various clusters.
LM 136 may generate an additional embedding VLM that characterizes the same data unit 304. LM 136 may be a natural language processing model. LM 136 may be trained to understand the meaning of tokens, data units, phrases, and/or the like. LM 136 may be or include an N-gram model, a unigram model, a bidirectional model, an exponential model, a neural network-based model, and/or other models, or any combination thereof. In some embodiments, LM 136 may be trained using data units labeled by SM 134 as ground truth. More specifically, a maximum aggregate score
for training data units may indicate the most likely cluster j. In some embodiments, the training data units may be data units previously processed by SM 134, e.g., as described above. In some embodiments, LM 136 may be initially pre-trained using some general vocabulary and then further trained using data units stored in various DPs 120-1, 120-2 . . . (with reference to
V
LM=(LM1, LM2, . . . LMM),
each component LMj being representative of the likelihood of the data unit 304 belonging to a corresponding cluster j.
Predictions of SM 134 and LM 136 may be aggregated (pooled) by a pooling classifier 330. In some embodiments, embeddings VLM and VSM may be aggregated using maximum pooling, where components of the pooled (aggregated) embedding V=(V1, V2, . . . . VM) are determined by the maximum values of the two components,
V
j=max (SMj, LMj).
In some embodiments, the components of the pooled embedding V may be determined as the average values of the two components,
In some embodiments, the components of the pooled embedding V may be determined as the weighted average values of the two components,
with weight w (e.g., defined in the range 0≤w≤1) determining a relative importance (trust) given to predictions SM 134 and LM 136; with weight w=½ indicating equal importance given to outputs of both models, weights w that are close to 0 indicating greater importance given to outputs of LM 136, and weights w that are close to 1 indicating greater importance given to outputs of SM 134.
Pooled embedding V=(V1, V2, . . . . VM) may be used as an input into classifier 340 that determines probabilities Pj that data unit 304 is associated with cluster j. In one illustrative non-limiting example, the probabilities may be computed using the softmax function,
with α being an empirically determined parameter. In some embodiments, α=1. Computed probabilities Pj may be used for final classification 350. Final classification 350 may include selecting a cluster, e.g., cluster k, associated with the highest probability Pk of the set of probabilities {Pj}. Once the maximum-probability cluster k has been identified, QPE 138 may use the identified cluster for retrieving data associated with that cluster. For example, final classification 350 of data unit 304 requesting information for “double pay” of employees of electric power stations may be associated with cluster “Overtime Pay.” Correspondingly, QPE 138 may retrieve all relevant data related to “overtime pay” category, e.g., using any suitable database search techniques.
In some implementations, final classification 350 may also include determining (and storing/outputting) a confidence level for the prediction. The confidence level may be determined by numerical value of the maximum probability Pk. In one illustrative example, the maximum probability Pk>0.8 may indicate a very high confidence level, the maximum probability in the range 0.6<Pk≤0.8 may indicate a high confidence level, the maximum probability in the range 0.4<Pk≤0.6 may indicate a medium confidence level, and the maximum probability in the range 0.2<Pk≤0.4 may indicate a low confidence level. It should be noted that these numbers are intended as an illustration only and that the thresholds separating various confidence levels may be set differently in various embodiments.
In some embodiments, confidence level may be determined in view of separate predictions made based on individual components of embeddings VSM and VLM output by the two models separately. For example, separate sets of probabilities {Pj−SM} and {Pj−LM} may be computed based on respective embeddings VSM and VLM, e.g., using a softmax classifier as described above or any other suitable classifier. In one example embodiment, a prediction may be determined to have a high confidence level provided that independent predictions of SM 134 and LM 136 coincide (the highest probabilities Pk−SM and Pk−LM correspond to the same cluster k) and the aggregated probability Pk is above a first empirically set threshold P1, e.g., P1=0.2, 0.3, 0.4, and/or the like. A prediction may also be determined to have a high confidence level if predictions of SM 134 and LM 136 do not coincide, but the higher prediction is above a second empirically set threshold P2. In some embodiments, P2>P1. If none of the above two scenarios is realized, the prediction may be determined to have a low confidence level.
At block 430, method 400 may include processing, by the processing device, the one or more tokens using a plurality of machine learning models (MLMs) to identify one or more clusters of a plurality of clusters. Each of the one or more identified clusters may be associated with at least one token of the one or more tokens of the data unit. In some embodiments, individual clusters of the plurality of clusters may be predefined by associating one or more anchor tokens (e.g., anchors 112 in
As illustrated with the callout portion of
At block 434, processing the one or more tokens may include using a second MLM (e.g., lexical model 136) of the plurality of MLMs. The second MLM may evaluate lexical associations of the one or more tokens with the plurality of clusters. In some embodiments, the second model may be trained (at least partially) using a plurality of training data units and ground truth labels generated by application of the first model to the plurality of training data units.
At block 440, method 400 may include retrieving, using the one or more identified clusters, data associated with the query. For example, having determined a semantic meaning of various tokens in the data unit, QPE 138 may identify and retrieve pertinent data stored in various data stores accessible to the processing device performing method 400.
In some embodiments, method 400 may further include updating one or more parameters of the first MLM in view of the one or more tokens and the one or more identified clusters. More specifically, the newly encountered tokens of the data unit may be used to further modify scores of the tokens based on the determined associations with the identified clusters, e.g., as described in conjunction with
At block 520, method 500 may continue with processing, using a first MLM, a training data unit that includes one or more tokens (e.g., tokens 204 in
At block 530, method 500 may include training a second MLM (e.g., lexical model 136) of the one or more MLMs using a plurality of training data units and ground truth labels generated by application of the first MLM to the plurality of training data units. In some embodiments, the training data units used for training of the second MLM may include at least some training data units that were used for training of the first MLM. In some embodiments, at least some of the training data units used for training of the second MLM may be different from the training data units that were used for training of the first MLM. In some embodiments, the second MLM may be an embeddings language model, e.g., nnlm-en-dim 128 model (or a similar model), that was pre-trained using English Google News 200B corpus (or a similar vocabulary).
The exemplary computer system 600 includes a processing device 602, a main memory 604 (e.g., read-only memory (ROM), flash memory, dynamic random access memory (DRAM) such as synchronous DRAM (SDRAM)), a static memory 606 (e.g., flash memory, static random access memory (SRAM)), and a data storage device 618, which communicate with each other via a bus 630.
Processing device 602 (which can include processing logic 603) represents one or more general-purpose processing devices such as a microprocessor, central processing unit, or the like. More particularly, the processing device 602 may be a complex instruction set computing (CISC) microprocessor, reduced instruction set computing (RISC) microprocessor, very long instruction word (VLIW) microprocessor, or a processor implementing other instruction sets or processors implementing a combination of instruction sets. The processing device 602 may also be one or more special-purpose processing devices such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), network processor, or the like. The processing device 602 is configured to execute instructions 622 for implementing query standardization engine 132 and/or training query processing engine 138 of
The computer system 600 may further include a network interface device 608. The computer system 600 also may include a video display unit 610 (e.g., a liquid crystal display (LCD) or a cathode ray tube (CRT)), an alphanumeric input device 612 (e.g., a keyboard), a cursor control device 614 (e.g., a mouse), and a signal generation device 616 (e.g., a speaker). In one illustrative example, the video display unit 610, the alphanumeric input device 612, and the cursor control device 614 may be combined into a single component or device (e.g., an LCD touch screen).
The data storage device 618 may include a computer-readable storage medium 624 on which is stored the instructions 622 embodying any one or more of the methodologies or functions described herein. The instructions 622 may also reside, completely or at least partially, within the main memory 604 and/or within the processing device 602 during execution thereof by the computer system 600, the main memory 604 and the processing device 602 also constituting computer-readable media. In some implementations, the instructions 622 may further be transmitted or received over a network 620 via the network interface device 608.
While the computer-readable storage medium 624 is shown in the illustrative examples to be a single medium, the term “computer-readable storage medium” should be taken to include a single medium or multiple media (e.g., a centralized or distributed database, and/or associated caches and servers) that store the one or more sets of instructions. The term “computer-readable storage medium” shall also be taken to include any medium that is capable of storing, encoding or carrying a set of instructions for execution by the machine and that cause the machine to perform any one or more of the methodologies of the present disclosure. The term “computer-readable storage medium” shall accordingly be taken to include, but not be limited to, solid-state memories, optical media, and magnetic media.
Although the operations of the methods herein are shown and described in a particular order, the order of the operations of each method may be altered so that certain operations may be performed in an inverse order or so that certain operations may be performed, at least in part, concurrently with other operations. In certain implementations, instructions or sub-operations of distinct operations may be in an intermittent and/or alternating manner.
It is to be understood that the above description is intended to be illustrative, and not restrictive. Many other implementations will be apparent to those of skill in the art upon reading and understanding the above description. The scope of the disclosure should, therefore, be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled.
In the above description, numerous details are set forth. It will be apparent, however, to one skilled in the art, that the aspects of the present disclosure may be practiced without these specific details. In some instances, well-known structures and devices are shown in block diagram form, rather than in detail, in order to avoid obscuring the present disclosure.
Some portions of the detailed descriptions above are presented in terms of algorithms and symbolic representations of operations on data bits within a computer memory. These algorithmic descriptions and representations are the means used by those skilled in the data processing arts to most effectively convey the substance of their work to others skilled in the art. An algorithm is here, and generally, conceived to be a self-consistent sequence of steps leading to a desired result. The steps are those requiring physical manipulations of physical quantities. Usually, though not necessarily, these quantities take the form of electrical or magnetic signals capable of being stored, transferred, combined, compared, and otherwise manipulated. It has proven convenient at times, principally for reasons of common usage, to refer to these signals as bits, values, elements, symbols, characters, terms, numbers, or the like.
It should be borne in mind, however, that all of these and similar terms are to be associated with the appropriate physical quantities and are merely convenient labels applied to these quantities. Unless specifically stated otherwise, as apparent from the following discussion, it is appreciated that throughout the description, discussions utilizing terms such as “receiving,” “determining,” “selecting,” “storing,” “analyzing,” or the like, refer to the action and processes of a computer system, or similar electronic computing device, that manipulates and transforms data represented as physical (electronic) quantities within the computer system's registers and memories into other data similarly represented as physical quantities within the computer system memories or registers or other such information storage, transmission or display devices.
The present disclosure also relates to an apparatus for performing the operations herein. This apparatus may be specially constructed for the required purposes, or it may comprise a general purpose computer selectively activated or reconfigured by a computer program stored in the computer. Such a computer program may be stored in a computer-readable storage medium, such as, but not limited to, any type of disk including floppy disks, optical disks, CD-ROMs, and magnetic-optical disks, read-only memories (ROMs), random access memories (RAMs), EPROMs, EEPROMs, magnetic or optical cards, or any type of media suitable for storing electronic instructions, each coupled to a computer system bus.
The algorithms and displays presented herein are not inherently related to any particular computer or other apparatus. Various general purpose systems may be used with programs in accordance with the teachings herein, or it may prove convenient to construct more specialized apparatus to perform the required method steps. The required structure for a variety of these systems will appear as set forth in the description. In addition, aspects of the present disclosure are not described with reference to any particular programming language. It will be appreciated that a variety of programming languages may be used to implement the teachings of the present disclosure as described herein.
Aspects of the present disclosure may be provided as a computer program product, or software, that may include a machine-readable medium having stored thereon instructions, which may be used to program a computer system (or other electronic devices) to perform a process according to the present disclosure. A machine-readable medium includes any mechanism for storing or transmitting information in a form readable by a machine (e.g., a computer). For example, a machine-readable (e.g., computer-readable) medium includes a machine (e.g., a computer) readable storage medium (e.g., read-only memory (“ROM”), random access memory (“RAM”), magnetic disk storage media, optical storage media, flash memory devices, etc.).
The words “example” or “exemplary” are used herein to mean serving as an example, instance, or illustration. Any aspect or design described herein as “example” or “exemplary” is not necessarily to be construed as preferred or advantageous over other aspects or designs. Rather, use of the words “example” or “exemplary” is intended to present concepts in a concrete fashion. As used in this application, the term “or” is intended to mean an inclusive “or” rather than an exclusive “or”. That is, unless specified otherwise, or clear from context, “X includes A or B” is intended to mean any of the natural inclusive permutations. That is, if X includes A; X includes B; or X includes both A and B, then “X includes A or B” is satisfied under any of the foregoing instances. In addition, the articles “a” and “an” as used in this application and the appended claims should generally be construed to mean “one or more” unless specified otherwise or clear from context to be directed to a singular form. Moreover, use of the term “an implementation” or “one implementation” or “an implementation” or “one implementation” throughout is not intended to mean the same implementation or implementation unless described as such. Furthermore, the terms “first,” “second,” “third,” “fourth,” etc. as used herein are meant as labels to distinguish among different elements and may not necessarily have an ordinal meaning according to their numerical designation.
Whereas many alterations and modifications of the disclosure will no doubt become apparent to a person of ordinary skill in the art after having read the foregoing description, it is to be understood that any particular implementation shown and described by way of illustration is in no way intended to be considered limiting. Therefore, references to details of various implementations are not intended to limit the scope of the claims, which in themselves recite only those features regarded as the disclosure.
