WO2011149608A1 - Identification et utilisation de champs critiques dans le cadre d'une gestion de qualité - Google Patents

Identification et utilisation de champs critiques dans le cadre d'une gestion de qualité Download PDF

Info

Publication number
WO2011149608A1
WO2011149608A1 PCT/US2011/033489 US2011033489W WO2011149608A1 WO 2011149608 A1 WO2011149608 A1 WO 2011149608A1 US 2011033489 W US2011033489 W US 2011033489W WO 2011149608 A1 WO2011149608 A1 WO 2011149608A1
Authority
WO
WIPO (PCT)
Prior art keywords
errors
error
operator
documents
operators
Prior art date
Application number
PCT/US2011/033489
Other languages
English (en)
Inventor
Arijit Sengupta
Brad A. Stronger
Original Assignee
Beyondcore, Inc.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from US12/787,342 external-priority patent/US7849062B1/en
Application filed by Beyondcore, Inc. filed Critical Beyondcore, Inc.
Publication of WO2011149608A1 publication Critical patent/WO2011149608A1/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/10—Office automation; Time management

Definitions

  • the present invention relates generally to quality management in a data-processing environment. Specifically, it relates to operational risk estimation and control associated with a data processing operation.
  • Errors in documents during a data processing operation for example, data entry and data transformation are common. These errors may result in significant losses to an organization, especially if a large amount of data is processed. It is therefore important to control the quality of documents.
  • Conventional techniques for controlling the quality of documents include error detection and correction, and determination of parameters for measuring errors.
  • One such measurement parameter can be the percentage of documents with errors. However, these parameters do not directly indicate the impact of the errors to the organization.
  • the conventional techniques for error detection are manual in nature. Errors can be detected by manually checking a set of documents to catch errors and compute the error rate. However, this technique may be error prone since the errors are detected manually. Further, the number of documents to be reviewed for catching errors (rather than just estimating error rates) is a function of the error rate. If the error rate is high, then a high percentage of documents need to be reviewed for catching a higher percentage of errors. Consequently, this technique can be labor intensive and therefore expensive.
  • Another technique for error prevention involves double typing the same document.
  • the two different versions of the same document are compared electronically, and any discrepancies are reviewed and corrected.
  • each document needs to be double typed, which can be a labor-intensive exercise.
  • the double typing and the confirmation of its correctness are done on a larger set of the documents.
  • a supervisor has to manually review each discrepancy to detect which of the two operators has made an error, or to correct the errors.
  • manual reviews themselves are prone to errors and result in wastage of labor, money and time.
  • Conventional techniques for detection of errors and correction are therefore cumbersome and expensive.
  • data entry operators can become aware as to when the supervisors are carrying out quality checks, and concentrate on quality for that period. If the process requires double entry of a complete document, it may result in 'gaming' of the system by the data entry operators, i.e., they may be lax in the initial data entry and catch errors if there is a discrepancy.
  • these techniques should enable benchmarking of organizations, i.e., how well organizations control data processing operational risk relative to one another. Such a benchmark should be comparable across process variations, organization size, document type, etc. Also, measurement schemes for data processing operators and systems should be directly correlated to measures used to evaluate the organizations. This enables true alignment of measurement schemes with performance requirements. These techniques should also deter 'gaming' of the system by data entry operators and supervisors.
  • Various embodiments of the invention provide methods and systems for identifying critical fields in documents, for example so that quality improvement efforts can be prioritized on the critical fields.
  • One aspect of the invention concerns a method for improving quality of a data processing operation in a plurality of documents.
  • a set of documents is sampled.
  • An error rate for fields in the documents is estimated based on the sampling.
  • Critical fields are identified based on which fields have error rates higher than a threshold. Which fields are the critical fields may be automatically updated on a dynamic basis.
  • the error rate for a field is based on both a frequency of errors in the field and a relative weight for that field. For example, the relative weight might be based on the operational impact of data processing errors in that field.
  • the threshold may be a predetermined constant value. Alternately, the threshold may vary as a function of the relative weight of a field. It may also be adjustable, either by the user or dynamically based on the sampled documents.
  • the threshold may be an aggregate across multiple fields, not just a threshold for a single field. For example, the set of critical fields may be determined by selecting the critical fields with the highest error rates until the aggregate sum of error rates reaches a threshold.
  • the threshold can also vary as a function of the distribution of error rates for the fields. For example, if the distribution of error rates is bimodal, the threshold may be set at some point between the two modes.
  • the error rate for a field is determined in part by estimating a probability that data entered for a field in a document is in error, without knowing a correct transcription for the field.
  • the data entered for a given field typically has a distribution among the different answers provided.
  • Data-entered answers that are identical form a cluster. For example, if three operators type (or otherwise data enter) the same answer for a field, that is a cluster.
  • a mode is the cluster for the most frequently appearing answer. There can be multiple modes if different answers are data-entered with the same frequency.
  • estimating the probability of error accounts for clusters, modes and/or their equivalencies.
  • Equivalencies can be determined based on the number of and sizes of clusters, as well as other factors.
  • the clusters that have the largest size for a field are determined to be equivalent and correct answers.
  • these clusters are determined to be not equivalent. Nevertheless, a single cluster is not selected as the correct answer. Rather, each non-equivalent cluster is assigned a probability of being a correct answer that is a function of the cluster's size.
  • the cluster, for which the associated operators have a lower average historical error rate is selected as a correct answer for a field.
  • Clusters could also be selected as the correct answer, based on whether the associated operators have a lower error rate for the field within the set of documents currently being evaluated or whether the associated operators have a lower historic error rate for the field.
  • Estimating the correct answer can also take into account whether the data entered for a field is the default value for that field.
  • Various embodiments of the present invention further provide methods and systems for quality management of a plurality of documents for a data-processing operation in an entity.
  • Each document comprises at least one field.
  • the entity includes an organization, or one or more employees of the organization.
  • the method measures the quality of a plurality of documents in a data-processing operation.
  • a relative operational risk is assigned for errors in each field of the plurality of documents. The assignment is based on the relative operational impact of the errors, and a frequency of errors is determined for each field.
  • an error rate is determined, based on the relative operational risk and the frequency of errors associated with each field.
  • a method for quality management of a plurality of documents for a data-processing operation in an entity comprises determination of error rates. Further, critical fields in the documents are dynamically identified based on the relative operational impact and the frequency of errors in the various fields. Errors are then reduced in the critical fields by using, for example, double typing of the data in the critical fields.
  • the occurrence of errors is predicted by determining a correlation between them and a set of process and external attributes.
  • the possibility of occurrence of the errors is notified to a supervisor if the attributes exhibit the characteristics correlated with errors.
  • the supervisor can then take preventive measures.
  • other preventative / corrective actions can be taken based on the predictions.
  • This process of error prediction, error rate computation and error prevention can be performed independently or iteratively, thereby reducing the occurrence of the errors.
  • the set of error correlation attributes and the set of critical fields also get updated depending upon changes in the measured error rate.
  • a set of documents is randomly identified for the purpose of sampling. Such a random sampling is used for determining the probability of errors related to specific fields of the documents.
  • the Operational risk weighted error' is identified for each employee for each field corresponding to the randomly sampled documents. This helps in identifying the specific training needs of the employees and in better targeting training efforts. Employees may also be assigned to various tasks based on their error rates.
  • a pattern of errors can be identified at a process level and an employee level.
  • the identified error patterns are then correlated with the root causes of errors.
  • a database is generated.
  • the database can then be used for identifying the root causes of further error patterns.
  • the database can be used to diagnose the root cause of an error pattern, for example, the root cause of an error pattern can be training related or process related or system related.
  • the database can also be used for a predictive diagnosis of the error.
  • the diagnosis may be a training, system or process error. If the diagnosis identifies a training need, then the method described in the previous paragraph can be used to better allocate training resources to the specific weaknesses of the employee or to specific weak employees. Employees may also be assigned to various tasks based on their error patterns.
  • the database can provide information regarding the historic diagnosis of previously observed error patterns corresponding to a field and/or an employee. For example, the database can provide historic data about diagnosis of a previous error or error pattern, and the methodology adopted at that time for mitigating the error.
  • the quality management system pertaining to the plurality of documents includes means for determining error rates.
  • the means for reducing errors is responsible for reducing errors by focusing on critical fields in the plurality of documents. It also updates the critical fields based on changes in error rates and patterns.
  • the means for predicting the occurrence of errors predicts errors by determining a correlation between the errors and a set of attributes. It also updates the set of attributes based on changes in error rates and patterns.
  • a means for controlling is used to coordinate between the remaining system elements of the quality management system. The means for controlling keeps a tab on the quality of the plurality of documents.
  • FIG. 1 is a block diagram illustrating an exemplary data-processing environment, suitable for use with the present invention
  • FIG. 2 is a flowchart depicting a method for measuring the quality of a plurality of documents in the data-processing environment, in accordance with an embodiment of the present invention
  • FIG. 3 is a flowchart depicting a method for reducing errors, in accordance with an embodiment of the present invention
  • FIG. 4 is a flowchart depicting a method for preventing errors, in accordance with an embodiment of the present invention.
  • FIG. 5 is a block diagram illustrating a system for quality management, in accordance with an embodiment of the present invention. DESCRIPTION OF PREFERRED EMBODIMENTS
  • Various embodiments of the present invention relate to quality management of an entity for a data-processing operation and provide methods and systems pertaining to operational risk control in the data-processing operations.
  • Data processing operations include, but are not limited to, data entry, transfer, storage, reporting and transformation.
  • the entity can be an organization such as a business process outsourcing organization or an in-house corporate data processing operation.
  • the entity can also be one or more employees of the organization.
  • Various embodiments of the invention measure error rate associated with a data processing operation for an employee or an organization. This involves identifying the relative operational impact associated with the errors and the frequency of the errors. Further, critical fields, i.e., the fields wherein the product of the relative operational impact of errors and error frequency can be large are identified.
  • critical fields are identified based on the frequency of errors and the relative operational impact of the errors in the fields. Data in these critical fields can be double typed to ensure that the errors in these critical fields are reduced. Subsequently, these critical fields can be updated and the process repeated on the new identified critical fields.
  • occurrences of errors are also predicted based on the correlation of errors with a set of attributes. Where a high correlation is identified between occurrence of errors and an attribute, a supervisor can be alerted regarding the same.
  • the supervisor can take preventive actions to avoid the occurrence of the errors.
  • other corrective / preventative measures can be undertaken.
  • the working of the error prediction process is verified by measuring the error rate.
  • the set of attributes is then updated based on the error rate.
  • FIG. 1 is a block diagram illustrating an exemplary data-processing environment that is suitable for use with various embodiments of the present invention.
  • the data-processing environment includes a process input block 102 that provides the input data, which is to be processed in the form of transcribed files or documents.
  • This input data is provided to employees 104, 106, and 108 in an organization.
  • the employee can then process the data, for example type in the data into electronic form.
  • Employees 104, 106, and 108 may be for example, medical transcription clerks, and data may be provided to them for medical transcription.
  • Only a few employees have been shown in FIG. 1. In actuality, the number of employees may be much higher.
  • the organization is a business process outsourcing (BPO) organization. While entering data, the employee may make errors.
  • a quality management block 110 controls the occurrence of errors in the document being processed.
  • quality management block 1 10 is responsible for detecting, preventing, predicting and controlling errors.
  • the processed documents are finally sent to a process output block 112 for delivery.
  • FIG. 2 is a flowchart depicting a method for measuring the quality of a plurality of documents for a data-processing operation, in accordance with an embodiment of the present invention. For the purpose of illustration, the method is hereinafter described assuming a data entry operation.
  • Each document can include several fields.
  • An exemplary document can include several fields such as 'Name', 'Address', 'Telephone Number', 'Email Address', 'Social Security Number', and so on.
  • an employee for example a data entry operator, can enter data in each of these fields.
  • some fields may be more important than others, for example, the social security number can be more important than the telephone number. Therefore, an error made while entering the social security number can have a greater impact or 'operational impact' than one made while entering the telephone number.
  • each field of a document can have a different operational impact.
  • a relative weight or 'relative operational risk' (w) is assigned to errors corresponding to each field of the plurality of documents at step 202.
  • Operational risk refers to the risk of losses due to errors in data processing operations.
  • Relative operational risk implies relative loss incurred due to errors in a field. The assignment is based on the operational impact of the errors, i.e., an error with a larger operational impact is weighted relatively higher than an error that has a smaller operational impact.
  • n a frequency (n) of errors is determined for each field in the plurality of documents, i.e., the number of errors in each field is determined.
  • n is determined by sampling a set of documents and measuring the number of errors in each field. Exemplary methods to determine n are described in the later part of the description section.
  • an error rate (E) is determined.
  • the error rate E is a measure of how well the operational risk is being controlled.
  • E is a measure of the quality of the plurality of documents and indicates the level of operational risk attributable to the data processing activities of an employee, a group of employees or the organization.
  • the error rate for a document across all the fields in the set S is given as
  • n is equal to zero or one for a given observation.
  • the average error rate for a sample of documents is given as i.e.,
  • EOR is the operational risk that is expected from a data processing operation. In an embodiment of the present invention, the EOR is obtained by multiplying e avg with the operational impact of making an error in every field in each of the plurality of documents.
  • EOR is a measure that can be used in accounting risk reserves and is relevant for regulations such as Sarbanes Oxley and Basel II. Consequently, E is directly related to how the organization as a whole is measured, thus effectively aligning measurement and performance across the layers of the organization.
  • Frequencies of errors in various fields are measured by sampling a set of documents from amongst the plurality of documents. Exemplary methods to sample the documents and identify the frequency of errors in the sampled documents are hereinafter described.
  • a set of documents of which the correct transcriptions (entries) are known a priori is sampled to estimate error rates.
  • a statistically significant sample size greater than 30 is considered.
  • the 95% confidence interval for the estimated error rate is sample mean ⁇ 2 x standard error of mean. It is to be noted that the sample size does not directly depend on the population size or the expected quality. However, the confidence interval could be tighter if the sample standard deviation is smaller.
  • one or more employees type the set of documents for which transcriptions already exist. This generates a new transcribed version of each document from the set of documents.
  • Each new transcription is then electronically compared with its corresponding known transcription, and any discrepancy between the two transcriptions is termed as an error, unless it has already been specified or learned (e.g., from cluster analysis). For example, if it is specified or learned that 'Lane' may also be typed as 'Ln.', this discrepancy is not considered to be an error.
  • n is recorded for each field in the plurality of documents. The recorded values of n are then used to determine E.
  • the E of a data entry operator is an absolute value, i.e., it is not relative to the error rates of other data entry operators.
  • a set of sections is identified in each document from amongst the set of documents for which transcriptions already exist.
  • a section may or may not include one or more fields.
  • Sections from amongst the set of sections are randomly combined to generate a new set of documents, wherein correct transcriptions are known for each component section.
  • the combination process is automated and ensures that each document in the new set of generated documents includes only one instance of each field in the original documents. In this way, a large number of documents with known transcriptions can be generated from a small number of documents with known transcriptions. For example, if there are m documents in the set of documents, and p sections in each document from amongst the set of documents, m p documents can be generated for sampling.
  • the new set of generated documents is then provided to employees for typing.
  • Each section of each new transcription is electronically compared with the corresponding section in the original set of documents with known transcriptions, and any discrepancy between the two transcriptions is termed as an error.
  • the E of a data entry operator is an absolute value, i.e., it is not relative to the error rates of other data entry operators.
  • a set of documents is identified randomly from amongst the plurality of documents for the purpose of sampling.
  • employees such as data processing operators are paired randomly, to generate a set of (one or more) random pairs of data entry operators.
  • the random pairs are generated such that no employee is in more than one pair for a given document.
  • a document from amongst the set of documents is typed by each data entry operator belonging to a pair from amongst the corresponding set of random pairs of data entry operators.
  • each document from amongst the set of documents is typed, so that there are at least two versions of each document. The two versions are electronically compared with each other, and any discrepancy is termed as an error.
  • n is recorded for each field in the plurality of documents.
  • the recorded values of n are then used to determine E.
  • E E of a data entry operator is relative to the error rates of other data entry operators. This is because a discrepancy could have been caused by either of the data processing operator in the random pair.
  • the error rates of multiple random samples are considered for computing the E of the data entry operator.
  • the sample can be large enough.
  • the random pairings of employees can be statistically 'backed out', i.e., average relative error rate of a specific employee can be measured.
  • the average relative frequency of errors for each field can also be accurately measured. This can be achieved without identifying the data entry operator or operators who have actually made the errors corresponding to the
  • This embodiment eliminates the need for sample documents with correctly known transcriptions. This embodiment is completely automatic and can be reliably carried out in a manner where employees can not become aware of which documents are being used for sampling and therefore can not 'game' the sampling.
  • a set of documents is identified randomly from the plurality of documents for the purpose of sampling.
  • employees such as data-processing operators are grouped randomly to generate one or more random groups of data-entry operators.
  • Each group includes at least three data-entry operators.
  • the random groups are so generated that no data-entry operator belongs to more than one group for a document.
  • Each data entry operator in a group types the same document from the set of given documents. In this way, each document from amongst the set of documents is typed, so that there are at least three versions of each document. These different versions of the same document are electronically compared with each other.
  • the most common answer is identified, based on the comparison.
  • the most common answer hereinafter referred to as the 'plurality vote' answer, is likely to be the correct answer as there are multiple ways to get an answer wrong, but only one way to get an answer right.
  • While identifying the plurality vote answer, 'specified equivalencies' or 'learned equivalencies' are also considered. For example, if it is specified that 'Lane' may also be typed as 'Ln.', both versions would be considered identical for the purposes of identifying the plurality vote answer. In some cases, more than one answer may appear equally often. If there are m different answers each occurring the same number of times, and no other answer occurring more frequently, referred to as multiple modes, each of these answers have equal probability of being the correct answer. The answers are assigned the probability of (m-l)/m of being an incorrect answer. Moreover, while assigning the probability of an incorrect answer, consideration can be taken of whether a multiple mode was the default value.
  • each mode has associated operators who have selected or data- entered that mode.
  • the historical error rate of the associated operators may be used to determine which mode is correct.
  • Other error rates may also be used, for example, the average error rate of the associated operators for the current field of interest, or averaged across the current set of documents.
  • this decrease in probability may not necessarily be proportional to the increase in the number of employees.
  • the plurality vote answers are combined, to automatically generate a plurality vote answer for the entire document.
  • Clusters are the same as modes, except that the size of each cluster need not be the same highest number. Modes are the clusters with the largest (same) size. Other factors can also be used to estimate likelihood of error based on comparisons of answers, without knowing a priori which answer is the correct answer. For example, if different employees data-enter different answers for the same field, some of these answers may be consistent with previously identified error patterns. These answers may be assigned a higher likelihood of being in error.
  • Each transcription entered by the employees is then electronically compared with its corresponding plurality vote answer. Any discrepancy between the two transcriptions is termed as an error, unless it has already been specified or learned. For example, if it is specified that 'Lane' may also be typed as 'Ln.', this discrepancy is not considered to be an error.
  • n is recorded for each field in the plurality of documents. The recorded values of n are then used to determine E. Accordingly, the value of E determined for a data entry operator is an absolute value, and is not relative to the error rates of other data entry operators.
  • various embodiments of the present invention eliminate the need for sampling documents with correctly known transcriptions. Such a process is completely automatic and can be carried out in a reliable manner wherein employees are not aware that they are being tested. This is because they have no way of differentiating the documents being used for sampling from the general flow of documents that they process. Therefore, an employee cannot 'game' the sampling.
  • w may be different for different fields. Further, an employee can make more errors in some fields of a document compared to other fields of the same document. Therefore, there can be critical fields in which the product of w and n are higher, compared to other fields.
  • the expected operational risk can be controlled by either controlling n or reducing w in the critical fields. Reducing errors in the critical fields can control the n. Changing operating processes can control the w.
  • critical fields within a document can be identified based on e;.
  • a small set of fields from amongst the complete set of fields can be the critical fields.
  • the employee/organization can substantially improve the overall quality, i.e., control the overall operational risk, by taking corrective/preventive actions in these critical fields. Since the corrective/preventive actions need to be taken in only a small set of fields, a greater proportion of expected operational risk can be avoided while incurring a proportionately lower cost.
  • FIG. 3 is a flowchart of the system for reducing errors, in accordance with an embodiment of the present invention.
  • a set of critical fields is identified in the plurality of documents. The identification of this set is based on w and n.
  • the error rate (e) of each field is determined, and the set of fields, of which the values of e are higher than a threshold, are identified as critical fields. For example, if 20% of the set of fields contribute 80% of the total error rate, then these 20% of the set of fields comprise the critical fields. In various embodiments of the invention, the identification of the critical fields can be automated.
  • the threshold is a predetermined value. In another embodiment of the present invention, the threshold depends on the operational impact and the value of n in each field in the plurality of documents. In another embodiment, the threshold is automatically set through standard applications such as 'goal seek' so that the sum of the e of the critical fields is equal to an arbitrary percentage (for example, 50%) of E (after accounting for any scaling factors).
  • the threshold is primarily set based on the customers' preference. Customers have to balance their risk tolerance and operational budget and decide their threshold for critical fields. The lower the risk appetite and the higher the operational budget, the greater is the percentage of document fields that can be considered critical fields. The distribution of errors among fields is also a factor determining the threshold. For example, if errors are highly concentrated among a few operationally important fields, then the threshold can be quite high (i.e. number of critical fields can be low) but still have the critical fields account for a significant proportion of expected operational risk.
  • the critical fields can also be similarly identified for each employee and training effort can be preferentially directed to the unique critical fields of each employee. This allows better targeting and customization and, therefore, better returns on investment of training activities.
  • the data in the identified set of critical fields are double typed for each critical field. In other words, two different data entry operators type the data in each of the identified sets of critical fields. The typing generates two different versions of the data in each critical field.
  • the two versions are compared with each other, and any discrepancy between the two is termed as an error.
  • the errors are then removed at step 308, to correct the data in the identified set of critical fields.
  • the errors can be removed through various ways.
  • a human supervisor may look into the error in order to mitigate it or the error may be removed by automatically calculating the plurality vote answer and then replacing the erroneous answer with the calculated plurality vote answer.
  • double typing 10 % of the fields in the plurality of documents can reduce E by 50%.
  • double typing a small number of fields reduces E by a large factor.
  • expending a small amount of labor and cost results in a large improvement in the quality of documents. Focusing on critical fields avoids undue usage of labor for error reduction.
  • the documents with corrected data can be subsequently sampled again at step 310 to check the value of E.
  • a decrease in E indicates a reduction in the operational risk.
  • an increase or no change in E indicates that critical fields may have changed.
  • Critical fields can change due to employee behavior or other changes in data entry operation.
  • the critical fields can be automatically updated, i.e., new critical fields are identified and the steps 302-310 repeated again.
  • the critical fields are also updated based on sampling. For example, companies may introduce process innovations that reduce error impacts or error frequency for the critical fields. As a result, the critical fields may shift. In such scenario, the critical fields are re-identified.
  • the correct typing is manually identified.
  • rules based or artificial intelligence algorithms can be used to identify the correct typing.
  • identifying the 'root cause' of errors can help prevent errors.
  • the root cause of errors may be determined by analyzing the error patterns in an automated or semi-automated manner.
  • the error identification and measurement procedures provide rich data on error patterns. For example, the Operational risk weighted error rate' for each employee for each data field can be easily identified. In some cases, a heavily skewed error pattern may be identified. In this case, for a given field, a small number of employees can have a disproportionately higher error rate than the average employees. This can indicate a training problem, which may be the root cause of these errors. In other cases, it may be found that almost all employees consistently make more errors in a particular field. This may indicate a process or system error.
  • a database of such error patterns and their corresponding historical diagnosis can be generated. Subsequently, the database can be used to automatically predict fields that may have clearly identifiable root causes of errors.
  • the database can additionally be used for diagnosing the possible cause of an error in that particular field.
  • the database can be used to automatically predict the fields that may have clearly identifiable root causes of errors and what may be the possible diagnosis for that field.
  • the possible diagnosis may be a training, system, or process error.
  • the database can be used to indicate what were the historic diagnoses and corresponding solutions in the database for the error pattern in question.
  • the prediction may be carried out using a simple correlation engine which identifies the most commonly occurring (or most highly correlated) root cause(s) for a given error pattern.
  • Other techniques such as more advanced clustering, pattern recognition and learning algorithms can be used to identify the appropriate cluster to which a specific error pattern belongs and what is the most likely diagnosis for that error pattern based on the database of previous error patterns and diagnoses.
  • FIG. 4 is a flowchart depicting the method for preventing errors by predicting the occurrence of errors (or by predicting an increase in the occurrence of errors).
  • a set of attributes is identified for correlation with the likelihood of occurrences of errors in the processed documents.
  • the attributes that are the best predictors of errors (most closely correlated with occurrences of errors) are identified.
  • a training process identifies the attributes.
  • the training is performed by using algorithms that measure correlation between an event (for example, an error) that has happened or not happened and an attribute (for example, the time of day). Other algorithms are based on artificial intelligence such as neural networks that use standard methodologies to identify such correlations.
  • data entry errors are mapped against various attributes to identify the best predictors of errors.
  • the occurrence of data entry errors can be mapped against the keystroke variability rate, i.e., the variation in the rate at which a user strokes the keys. It is observed that the frequency of errors increases with increase in the keystroke variability rate. Therefore, keystroke rate variability can be a good attribute for error prediction.
  • the occurrence of data entry errors is mapped against several other attributes to determine the attributes that are the best predictors of errors.
  • an exemplary learning algorithm is selected to ensure best prediction of errors based on the identified attributes.
  • Step 406 may alternatively be performed before step 404, i.e., a best predictive algorithm is first identified and then the algorithm is used in training mode to identify the best predictive attributes.
  • the correlation is determined between the errors in the plurality of documents and a set of attributes. This correlation is based on the identified learning algorithm.
  • the learning algorithm can be based on for example, fuzzy logic, neural network, Bayes Nets, abstract local search and genetic algorithm.
  • a learning algorithm can establish a correlation between two events, for example, for two given events A and B.
  • the learning algorithm can establish that if A occurs, it is likely that B also do so.
  • the learning algorithm can learn which attributes have the strongest correlation with, or are the best indicators of the occurrence of errors.
  • Exemplary attributes can be the lighting conditions in the data entry operations, the complexity of the document being processed, the eye color of the data entry operator, the time when the errors were made, backlog levels when the errors occurred, and the variability of the keystroke rate of the data entry operator when the errors occurred.
  • the learning algorithm can determine that the keystroke rate variability is a good indicator of the occurrence of errors. This correlation can now be used to predict the occurrence of errors.
  • the learning algorithm can also determine that the eye color of the data entry operator is not correlated with him or her making errors. Therefore, the learning algorithm will reject this attribute.
  • the supervisor is appropriately notified about the likelihood of errors occurring. For example, if the keystroke rate of a data entry operator shows high variations, a supervisor of the data entry operator can be notified that it is likely that the data entry operator may make an error in the near future.
  • the supervisor can then take preventive actions to prevent errors. For example, the supervisor can verify the prediction by checking the data entered by the operator. Further, the supervisor can alert the data entry operator if errors are identified. The supervisor may also offer the data entry operator a short break or reduce his or her backlog levels.
  • the system may initiate alternative preventative / corrective actions such as routing data for double typing. For example, the system can ask another employee to double type the data. It is to be noted these corrective and preventive actions are exemplary and any other corrective/preventive action can be taken without diverting from the scope and spirit of the invention.
  • the error rate is monitored to confirm that the error prediction process has not gone out of synch.
  • the plurality of documents is periodically sampled to monitor the error prediction process, and E is determined subsequently. Monitoring is required to check the correctness of the error prediction process, for example, the learning algorithm may predict that a particular operator is going to make errors. However, the next few documents typed by him may contain no errors. Such inconsistencies in error prediction can be verified, based on the value of E. For example, a low value of E in the above-mentioned case can imply that the learning algorithm has gone out of calibration. This is because the operator may adapt his behavior accordingly, for example, errors may occur when operators chat among themselves and stop typing while processing a document.
  • the supervisor is notified that errors are likely to occur.
  • the supervisor checks on the operators.
  • the operators may realize that the supervisor checks on them whenever they start chatting, and therefore stop chatting among themselves. This, in turn, can prevent the occurrence of errors due to chatting.
  • errors may now occur due to other attributes not known to the learning algorithm.
  • the learning algorithm is recalibrated. This recalibration may be initiated automatically or manually and can be achieved by updating the set of attributes, i.e., by identifying new attributes that are likely to cause errors and rejecting those that are not correlated to errors; and/or by selecting a better prediction algorithm as described in steps 404, and 406.
  • the error measurement algorithms described above such as the plurality vote algorithm, generate rich data on the specific error patterns of each data entry employee. Such data can be used to double check the data entered by an employee. For example, an employee may have the habit of typing '7' instead of the character 'Z.' Such error patterns are highly employee-specific and generic rules to catch such errors may not be very effective.
  • the employee-specific error patterns gathered through the error measurement algorithms can be used to customize deterministic algorithms specific to each employee (e.g., employee-specific rules), or to train learning algorithms specific to each employee. This specificity can significantly increase the effectiveness of such algorithms. This specificity can also be applied on a field by field basis, for example to generate field-specific rules or to train learning algorithms specific to each field.
  • the quality of the plurality of documents is managed in an embodiment of the present invention.
  • E is measured to check the initial quality of the plurality of documents.
  • the errors are then reduced, as described earlier.
  • the occurrence of errors may also be prevented by identifying and mitigating 'root causes' of errors or by predicting such errors.
  • the process of measuring E, and reducing errors can be performed repetitively to monitor and control the overall quality of the documents generated by the employee. It should be noted that the error measurement, reduction and prediction processes could operate independently. They can also operate simultaneously or at different times. These processes can make use of one or more sampling schemes, described earlier, to measure E. They can also use any other sampling scheme without limiting the scope of the present invention.
  • FIG. 5 is a block diagram illustrating quality management system 500, in accordance with an embodiment of the present invention.
  • Quality management system 500 includes an error rate measurement module 502, an error reduction module 504, an error occurrence prediction module 506, and a control module 508.
  • Error rate measurement module 502 is the means for determining E; error reduction module 504 enables reduction of the errors in the critical fields of the plurality of documents; and error occurrence prediction module 506 prevents errors by predicting their occurrence, and establishes a correlation between errors and a set of attributes by implementing learning algorithms.
  • Control module 508 coordinates the other modules of the software system to control the quality of the plurality of documents.
  • control module 508 monitors the change in the error rates on account of preventive/corrective actions taken to reduce the errors.
  • Control module 508 updates the set of attributes for module 506 in case the attributes that impact the error occurrences change. Further, it periodically updates the critical fields for module 504. For example, companies may introduce process innovations that reduce error impacts or error frequency for the initially identified critical fields. Consequently, the critical fields can shift.
  • system elements of quality management system are examples of quality management system.
  • An exemplary data reporting operation can be an advance shipment note that is sent by a client to a recipient of the shipment as well as to the shipment agency, for example a courier agency. There can be discrepancies in the shipment notes send to the recipient and the shipment agency.
  • the various embodiments of the invention can be used to measure the quality of reporting of shipment details by the client.
  • the error- identification technology or the plurality vote answer generation algorithm can be used to improve the error rate of Optical Character Recognition (OCR) systems.
  • OCR Optical Character Recognition
  • the same document can be scanned by three or more different OCR systems, in order to automatically generate a plurality vote answer from the output of the OCRs.
  • This plurality vote answer is likely to be more accurate than any of the individual OCR scans.
  • the error measurement algorithms such as those based on the plurality vote answer generation algorithm, can also be used to quickly measure the operational risk due to differences in systems that are supposed to have identical output. For example, a bank may acquire another bank and wish to merge their existing systems. A random statistical sampling could be carried out with a representative sample, and the operational risk measure E could be used to quantify the discrepancies between the disparate systems that have to be consolidated.
  • the underlying patterns of the differences between the original process and the transferred process can be used to direct documentation efforts to the specific parts of the original process that seem to be ambiguous, direct potentially operator-specific training efforts to the parts of the transferred process that require the most training, and automation efforts to parts of the process that can be automated based on the patterns observed.
  • These actions can be based on the measured errors, measured error rates (which accounts for both frequency of errors and relative operation risk) and/or error patterns (including patterns in the differences between the data-entered or the errors for the original and transferred processes).
  • the embodiments of the present invention have the advantage that they provide an error rate that directly measures the effectiveness in controlling the operational risk of an organization or employee corresponding to a data processing operation.
  • the error rate can also be used to measure the expected operational risk of the data processing operation, thus it is useful for setting up accounting risk reserves and for meeting regulatory requirements (or other operational risk requirements) such as Sarbanes Oxley and Basel II.
  • the embodiments of the invention also allow rating/benchmarking of organizations and employees on the basis of how well they control operational risk, thus enabling an apples-to- apples comparison between organizations with different processes, document structure, size, etc.
  • the embodiments of the present invention offer a predominantly or completely automated method and system for reduction, prevention and prediction of errors in data processing operations.
  • the various embodiments allow avoiding a large percentage of expected operational risk while expending a relatively small amount of labor. This is achieved by systematically focusing on the critical fields of the document, which accounts for a
  • Various embodiments of the present invention eliminate the need for sampling documents with known correct transcriptions. Such a process is completely automatic and can be reliably carried out in a manner where employees are not aware that they are being tested. This is because they have no way of differentiating the documents being used for sampling from the general flow of documents that they process. Therefore, an employee cannot 'game' the sampling.
  • inventions provide a method for identifying critical fields for each employee. Therefore, training effort can be directed toward the critical fields identified for each employee. This allows tailored targeting/customization, thereby ensuring better returns on investment of training activities. Error rates can also be estimated without identifying which specific operator was responsible for a discrepancy. Error rate estimation can be achieved by sampling a small number of documents.
  • the quality management can be performed real time. Further, since the process is automated, the quality management can be performed real time. Further, the employees need not be aware that their quality is being audited. Further, unlike in training intensive procedures such as "Six Sigma" the data entry operators do not need to be specifically trained to use these methodologies which may be automated.
  • the system may be embodied in the form of a computer system.
  • Typical examples of a computer system include a general-purpose computer, a programmed microprocessor, a micro-controller, a peripheral integrated circuit element, and other devices or arrangements of devices that are capable of implementing the steps that constitute the method of the present invention.
  • the computer system comprises a computer, an input device, a display unit and the Internet.
  • the computer comprises a microprocessor.
  • the microprocessor can be one or more general- or special-purpose processors such as a Pentium®, Centrino®, Power PC®, and a digital signal processor.
  • the microprocessor is connected to a communication bus.
  • the computer also includes a memory, which may include Random Access Memory (RAM) and Read Only Memory (ROM).
  • RAM Random Access Memory
  • ROM Read Only Memory
  • the computer system also comprises a storage device, which can be a hard disk drive or a removable storage device such as a floppy disk drive, optical disk drive, and so forth.
  • the storage device can also be other similar means for loading computer programs or other instructions into the computer system.
  • the computer system also includes one or more user input devices such as a mouse and a keyboard, and one or more output devices such as a display unit and speakers.
  • the computer system includes an operating system (OS), such as Windows, Windows CE, Mac, Linux, Unix, a cellular phone OS, or a proprietary OS.
  • OS operating system
  • the computer system executes a set of instructions that are stored in one or more storage elements, to process input data.
  • the storage elements may also hold data or other information as desired.
  • a storage element may be an information source or physical memory element present in the processing machine.
  • the set of instructions may include various commands that instruct the processing machine to perform specific tasks such as the steps that constitute the method of the present invention.
  • the set of instructions may be in the form of a software program.
  • the software may be in various forms, such as system software or application software. Further, the software may be in the form of a collection of separate programs, a program module with a larger program, or a portion of a program module.
  • the software might also include modular programming in the form of object-oriented programming and may use any suitable language such as C, C++ and Java.
  • the processing of input data by the processing machine may be in response to user commands to results of previous processing, or in response to a request made by another processing machine. While the preferred embodiments of the invention have been illustrated and described, it will be clear that it is not limited to these embodiments only. Numerous modifications, changes, variations, substitutions and equivalents will be apparent to those skilled in the art, without departing from the spirit and scope of the invention, as described in the claims.

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Strategic Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Human Resources & Organizations (AREA)
  • Operations Research (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Data Mining & Analysis (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Debugging And Monitoring (AREA)
  • Complex Calculations (AREA)

Abstract

La présente invention concerne des procédés et des systèmes d'identification de champs critiques dans des documents, par exemple pour que des efforts d'amélioration de la qualité puissent être priorisés sur les champs critiques. Un aspect de l'invention concerne un procédé d'amélioration de la qualité d'une opération de traitement de données dans une pluralité de documents. Un ensemble de documents est échantillonné. Un taux d'erreur des champs dans les documents est estimé sur la base de l'échantillonnage. Les champs critiques sont identifiés sur la base des champs présentant un taux d'erreur supérieur à un seuil.
PCT/US2011/033489 2010-05-25 2011-04-21 Identification et utilisation de champs critiques dans le cadre d'une gestion de qualité WO2011149608A1 (fr)

Applications Claiming Priority (12)

Application Number Priority Date Filing Date Title
US12/787,342 2010-05-25
US12/787,342 US7849062B1 (en) 2005-03-18 2010-05-25 Identifying and using critical fields in quality management
US12/944,554 US20110055620A1 (en) 2005-03-18 2010-11-11 Identifying and Predicting Errors and Root Causes in a Data Processing Operation
US12/944,547 2010-11-11
US12/944,541 US8019734B2 (en) 2005-03-18 2010-11-11 Statistical determination of operator error
US12/944,541 2010-11-11
US12/944,559 2010-11-11
US12/944,559 US7933878B2 (en) 2005-03-18 2010-11-11 Assessing and managing operational risk in organizational operations
US12/944,547 US7933934B2 (en) 2005-03-18 2010-11-11 Operator-specific quality management and quality improvement
US12/944,554 2010-11-11
US12/945,754 2010-11-12
US12/945,754 US20110066901A1 (en) 2005-03-18 2010-11-12 Automated Quality Management System

Publications (1)

Publication Number Publication Date
WO2011149608A1 true WO2011149608A1 (fr) 2011-12-01

Family

ID=45004270

Family Applications (2)

Application Number Title Priority Date Filing Date
PCT/US2011/033489 WO2011149608A1 (fr) 2010-05-25 2011-04-21 Identification et utilisation de champs critiques dans le cadre d'une gestion de qualité
PCT/US2011/037956 WO2011150097A2 (fr) 2005-03-18 2011-05-25 Identification et utilisation de champs critiques dans une gestion de qualité

Family Applications After (1)

Application Number Title Priority Date Filing Date
PCT/US2011/037956 WO2011150097A2 (fr) 2005-03-18 2011-05-25 Identification et utilisation de champs critiques dans une gestion de qualité

Country Status (2)

Country Link
GB (1) GB2498440A (fr)
WO (2) WO2011149608A1 (fr)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8782087B2 (en) 2005-03-18 2014-07-15 Beyondcore, Inc. Analyzing large data sets to find deviation patterns
US10127130B2 (en) 2005-03-18 2018-11-13 Salesforce.Com Identifying contributors that explain differences between a data set and a subset of the data set
US10140422B2 (en) 2013-03-15 2018-11-27 Battelle Memorial Institute Progression analytics system
US10796232B2 (en) 2011-12-04 2020-10-06 Salesforce.Com, Inc. Explaining differences between predicted outcomes and actual outcomes of a process
US10802687B2 (en) 2011-12-04 2020-10-13 Salesforce.Com, Inc. Displaying differences between different data sets of a process

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20030046318A1 (en) * 2001-05-08 2003-03-06 Schohn Gregory C. Reorganizing content of an electronic document
US6738798B1 (en) * 2000-06-01 2004-05-18 Ge Medical Technology Services, Inc. Automated monitoring of collection of operational data from medical imaging devices
US7720822B1 (en) * 2005-03-18 2010-05-18 Beyondcore, Inc. Quality management in a data-processing environment

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5251131A (en) * 1991-07-31 1993-10-05 Thinking Machines Corporation Classification of data records by comparison of records to a training database using probability weights
US5488725A (en) * 1991-10-08 1996-01-30 West Publishing Company System of document representation retrieval by successive iterated probability sampling
AUPR033800A0 (en) * 2000-09-25 2000-10-19 Telstra R & D Management Pty Ltd A document categorisation system
US8131756B2 (en) * 2006-06-21 2012-03-06 Carus Alwin B Apparatus, system and method for developing tools to process natural language text
US7849398B2 (en) * 2007-04-26 2010-12-07 Xerox Corporation Decision criteria for automated form population

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6738798B1 (en) * 2000-06-01 2004-05-18 Ge Medical Technology Services, Inc. Automated monitoring of collection of operational data from medical imaging devices
US20030046318A1 (en) * 2001-05-08 2003-03-06 Schohn Gregory C. Reorganizing content of an electronic document
US7720822B1 (en) * 2005-03-18 2010-05-18 Beyondcore, Inc. Quality management in a data-processing environment

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8782087B2 (en) 2005-03-18 2014-07-15 Beyondcore, Inc. Analyzing large data sets to find deviation patterns
US9390121B2 (en) 2005-03-18 2016-07-12 Beyondcore, Inc. Analyzing large data sets to find deviation patterns
US10127130B2 (en) 2005-03-18 2018-11-13 Salesforce.Com Identifying contributors that explain differences between a data set and a subset of the data set
US10796232B2 (en) 2011-12-04 2020-10-06 Salesforce.Com, Inc. Explaining differences between predicted outcomes and actual outcomes of a process
US10802687B2 (en) 2011-12-04 2020-10-13 Salesforce.Com, Inc. Displaying differences between different data sets of a process
US10140422B2 (en) 2013-03-15 2018-11-27 Battelle Memorial Institute Progression analytics system
US10872131B2 (en) 2013-03-15 2020-12-22 Battelle Memorial Institute Progression analytics system

Also Published As

Publication number Publication date
WO2011150097A3 (fr) 2012-02-09
GB201223364D0 (en) 2013-02-06
WO2011150097A2 (fr) 2011-12-01
GB2498440A (en) 2013-07-17

Similar Documents

Publication Publication Date Title
US7849062B1 (en) Identifying and using critical fields in quality management
US7925638B2 (en) Quality management in a data-processing environment
US7720822B1 (en) Quality management in a data-processing environment
US9390121B2 (en) Analyzing large data sets to find deviation patterns
US11282035B2 (en) Process orchestration
US20150254791A1 (en) Quality control calculator for document review
JP2012504289A (ja) オンラインビジネストランザクションデータを使用するローンアクセスの評価
JP2008533623A (ja) リスクに基づくデータ評価
US8756071B2 (en) Methods and apparatus for queue-based cluster analysis
WO2011149608A1 (fr) Identification et utilisation de champs critiques dans le cadre d'une gestion de qualité
WO2016084642A1 (fr) Serveur, système et programme d'examen de crédit
US20190172069A1 (en) Computer-based Understanding of Customer Behavior Patterns for Better Customer Outcomes
CN109214912A (zh) 行为数据的处理方法、行为预测方法、装置、设备及介质
US20160253752A1 (en) Method and system for internal analysis of loan instruments
CN117540826A (zh) 机器学习模型的优化方法、装置、电子设备及存储介质
US20170270611A1 (en) Processing system to automatically assign electronic records to verification process subset levels
CN108197740A (zh) 企业倒闭预测方法、电子设备和计算机存储介质
CN115562981A (zh) 基于机器学习的软件质量评估方法
US20220101061A1 (en) Automatically identifying and generating machine learning prediction models for data input fields
EP4248292A1 (fr) Systèmes et procédés pour des prédictions de plage de règlement de transaction d'intervalle de confiance
CN115485662A (zh) 计算平台上的配额请求解析
US20230394069A1 (en) Method and apparatus for measuring material risk in a data set
US20220027831A1 (en) System and method for security analyst modeling and management
US20240020408A1 (en) Masking compliance measurement system
CN117077189A (zh) 银行业威胁建模平台建设方法、设备及可读存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 11787070

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 11787070

Country of ref document: EP

Kind code of ref document: A1