Systems and methods for conversion of administrative systems
Methods, systems and apparatuses, including computer programs encoded on computer storage media, are provided for automatically converting a legacy administration system to a target administration system. The data definitions, data items, data sets, product features, business functions/processes, business rules, and calculation modules of both systems are analyzed by an analysis system to determine elements in common. Common elements between the systems. are determined using AI techniques such as natural language processing (NLP). Elements of the legacy system not present in the target system are generated and the legacy system is then migrated to the target system. For data required by the target system not present in the legacy system, data-driven prediction models may be used to predict the required data.
1. A method for migrating a part of a legacy administration system, the legacy administration system comprising legacy data items, legacy functionality, and legacy product types, to a target administration system, the target administration system comprising target data items, target functionality, and target product types, each legacy data item and target data item comprising at least one data type, the method comprising:
comparing a legacy data item with each target data item of the same or similar type and calculating a similarity score for each comparison;
identifying a matching target data item when the similarity score is greater than or equal to a threshold value;
creating a new data item in the target administration system corresponding to the legacy data item when a matching data item is not found;
comparing a legacy functionality with each target functionality and calculating a similarity score for each comparison;
identifying a matching target functionality when the similarity score is greater than or equal to a threshold value; and
creating a new functionality in the target administration system corresponding to the legacy functionality when a matching functionality is not found;
thereby migrating the legacy data item and the legacy functionality from the legacy administration system to the target administration system.
2. The method of claim 1 , wherein the comparing a legacy data item with each target data item comprises using natural language processing (NLP).
3. The method of claim 2 , wherein the comparing a legacy data item with each target data item comprises:
tokenizing and stemming a description associated with the legacy data item;
tokenizing and stemming descriptions associated with the target data items;
vectorizing the tokenized and stemmed legacy data item description;
vectorizing the tokenized and stemmed target data item descriptions; and
calculating a similarity score between the vectorized legacy data item description and each vectorized target data item description.
4. The method of claim 3 , wherein at least one data item description is extracted from a data source using image processing and optical character recognition (OCR).
5. The method of claim 1 , wherein the comparing a legacy functionality with each target functionality comprises using natural language processing (NLP).
6. The method of claim 5 , wherein the comparing a legacy functionality with each target functionality comprises:
tokenizing and stemming a description associated with the legacy functionality;
tokenizing and stemming descriptions associated with the target functionalities;
vectorizing the tokenized and stemmed legacy functionality description;
vectorizing the tokenized and stemmed target functionality descriptions;
calculating a description similarity score between the vectorized legacy functionality description and each vectorized target functionality description.
7. The method of claim 6 , wherein at least one functionality description is extracted from a data source.
8. The method of claim 6 , wherein the legacy functionality comprises a set of inputs and a set of outputs, each target functionality comprises a set of inputs and a set of outputs, and wherein the comparing a legacy functionality with each target functionality further comprises comparing the set of inputs of the legacy functionality with the set of inputs of each target functionality and comparing the set of outputs of the legacy functionality with the set of outputs of each target functionality to determine a data similarity score for each comparison.
9. The method of claim 8 , wherein the description similarity score and the data similarity scores for each comparison are combined to generate an overall similarity score for each function comparison.
10. The method of claim 1 , further comprising:
identifying data required by the target system not present in the legacy system;
generating a prediction model targeting the required data and trained using historical data or business rules in the legacy system;
predicting the required data using the prediction model; and
storing the predicted data in the target system.
11. The method of claim 1 , further comprising generating a product type mapping between the legacy system and the target system, the generating comprising:
identifying at least one associated policy form for each product type in the legacy system;
identifying at least one associated policy form for each product type in the target system;
extracting features and associated descriptions from the identified forms using image processing and NLP;
for each product feature in the legacy system, calculating similarity scores between the legacy feature and each feature in the target system using the associated descriptions; and
for each product in the legacy system, calculating a combined similarity score between the legacy product and each product in the target system using the feature similarity scores.
12. A system for migrating a part of a legacy administration system, the legacy administration system comprising legacy data items and legacy functionality, to a target administration system, the target administration system comprising target data items and target functionality, each legacy data item and target data item comprising at least one data type, the system comprising:
a data mapping system adapted to compare a legacy data item with each target data item of the same or similar type and calculate a similarity score for each comparison;
a function mapping system adapted to compare a legacy functionality with each target functionality and calculate a similarity score for each comparison;
a product mapping system adapted to compare a legacy product with each target product to identify a target product that matches the legacy product;
a migration system adapted to:
convert at least one data item from the legacy administration system to the target administration system using the similarity scores calculated for the data item comparisons;
convert at least one functionality from the legacy administration system to the target administration system using the similarity scores calculated for the functionality comparisons; and
convert at least one product from the legacy administration system to the target administration system using the similarity scores calculated for the product comparisons.
13. The system of claim 12 , wherein the data mapping system is adapted to compare a legacy data item with each target data item using natural language processing (NLP).
14. The system of claim 13 , wherein the comparing a legacy data item with each target data item comprises:
tokenizing and stemming a description associated with the legacy data item;
tokenizing and stemming descriptions associated with the target data items;
vectorizing the tokenized and stemmed legacy data item description;
vectorizing the tokenized and stemmed target data item descriptions;
calculating a similarity score between the vectorized legacy data item description and each vectorized target data item description.
15. The system of claim 14 , wherein at least one data item description is extracted from a data source using a data processing system using image processing and optical character recognition (OCR).
16. The system of claim 12 , wherein the function mapping system is adapted to compare a legacy functionality with each target functionality using natural language processing (NLP).
17. The system of claim 16 , wherein the comparing a legacy functionality with each target functionality comprises:
tokenizing and stemming a description associated with the legacy functionality;
tokenizing and stemming descriptions associated with the target functionalities;
vectorizing the tokenized and stemmed legacy functionality description;
vectorizing the tokenized and stemmed target functionality descriptions;
calculating a similarity score between the vectorized legacy functionality description and each vectorized target functionality description.
18. The system of claim 17 , wherein at least one functionality description is extracted from a data source using image processing and optical character recognition (OCR).
19. The system of claim 17 , wherein the legacy functionality comprises a set of inputs and a set of outputs, each target functionality comprises a set of inputs and a set of outputs, and wherein the comparing a legacy functionality with each target functionality further comprises comparing the set of inputs of the legacy functionality with the set of inputs of each target functionality and comparing the set of outputs of the legacy functionality with the set of outputs of each target functionality to determine a data similarity score for each comparison.
20. The system of claim 19 , wherein the function mapping system is adapted to combine the description similarity score and the data similarity score for each comparison to generate an overall similarity score for each comparison.
21. The system of claim 12 , further comprising a data modeling system, and wherein the migration system is configured to:
identify data required by the target system not present in the legacy system;
generate a prediction model using the data modeling system, the prediction model targeting the required data and trained using historical data in the legacy system;
predict the required data using the prediction model; and
store the predicted data in the target administration system.
22. The system of claim 12 , wherein the product mapping system is configured to generate a product type mapping between the legacy system and the target system, the generating comprising:
identifying at least one associated policy form for each product type in the legacy system;
identifying at least one associated policy form for each product type in the target system;
extracting features and associated descriptions from the identified forms using image processing and NLP;
for each product feature in the legacy system, calculating similarity scores between the legacy feature and each feature in the target system using the associated descriptions; and
for each product in the legacy system, calculating a combined similarity score between the legacy product and each product in the target system using the feature similarity scores.