Method and system for machine learning based integration of bots
Developing a new solution or revising an existing solution to meet organizational requirements is often time consuming. One solution to overcome the above problem is to integrate existing solution providers like bots without affecting the overall performance. Conventional methods integrate bots performing similar operations. To overcome the challenges in the conventional approaches, the present disclosure provides a method and system for Machine learning (ML) based integration of bots. The present disclosure integrates a plurality of dissimilar bots using ML approach. Further, the present disclosure enables an end-user to combine two or more similar or dissimilar bots using an interface. The end-user/user can create, combine, validate and test the combined dissimilar bots without any prior knowledge of programming. Furthermore, the present disclosure auto-recommends the end-user regarding which bots can be combined with the selected bots by checking the technical and end-user feasibility.
1 . A processor implemented method, the method comprising:
receiving, by one or more hardware processors, at least two bots selected by a user from a bot repository as input, wherein each bot comprises a plurality of bot features;
computing, by the one or more hardware processors, an end-user bot compatibility value for the at least two bots based on a corresponding plurality of bot features using a trained classifier, wherein the end-user compatible value is set to one only if the corresponding end-user bot compatibility score is greater than a predefined threshold;
generating, by the one or more hardware processors, a technical compatibility value of the at least two bots based on a plurality of input features and a plurality of output features associated with the at least two bots, wherein the technical compatibility value is set to one only if the plurality of output features of a first bot from among the at least two bots is equal to the plurality of input features associated with the second bot from among the at least two bots;
updating, by the one or more hardware processors, a dynamic compatibility matrix with the at least two bots based on the end-user bot compatibility value and the technical compatibility value, wherein the dynamic compatibility matrix is updated with a value one if the end-user compatibility value and the technical feasibility value are set to one and zero otherwise; and
creating, by the one or more hardware processors, a bot package by integrating the at least two bots based on the dynamic compatibility matrix, wherein the bot package is created only if a corresponding entry between the at least two bots in the dynamic compatibility matrix is set to one.
2 . The processor implemented method of claim 1 , wherein the method of training the classifier for computing the end-user compatibility value of the at least two bots comprises:
receiving a plurality of bot features corresponding to each of a plurality of bots stored in the bot repository;
generating a preprocessed plurality of bot features corresponding to each of the plurality of bots using a preprocessing technique;
generating a plurality of vectorized bot features associated with each of the plurality of bots based on a corresponding preprocessed plurality of bot features using a vectorization technique;
generating a plurality of bot combinations based on the plurality of bots using a permutation combination technique; and
training the classifier based on the plurality of vectorized bot features associated with each of the plurality of bots and the plurality of bot combinations, wherein the classifier is trained to compute a compatibility score between each of the plurality of bot combinations.
3 . The processor implemented method of claim 1 , wherein the plurality of bot features comprises a bot name, the plurality of input features, the plurality of output features, a datatype associated with each of the plurality of input features, a datatype associated with each of the plurality of output features and a bot category.
4 . The processor implemented method of claim 1 , further comprises auto recommending a plurality of bots to the user using a Graph Neural Network (GNN) based self-learning model, wherein the GNN based self-learning model is associated with the dynamic compatibility matrix further comprising compatibility values of the plurality of bots.
5 . The processor implemented method of claim 1 , wherein the created bot package is included in the bot repository.
6 . A system comprising:
at least one memory storing programmed instructions; one or more Input/Output (I/O) interfaces; and one or more hardware processors operatively coupled to the at least one memory, wherein the one or more hardware processors are configured by the programmed instructions to:
receive at least two bots selected by a user from a bot repository as input, wherein each bot comprises a plurality of bot features;
compute an end-user bot compatibility value for the at least two bots based on a corresponding plurality of bot features using a trained classifier, wherein the end-user compatible value is set to one only if the corresponding end-user bot compatibility score is greater than a predefined threshold;
generate a technical compatibility value of the at least two bots based on a plurality of input features and a plurality of output features associated with the at least two bots, wherein the technical compatibility value is set to one only if the plurality of output features of a first bot from among the at least two bots is equal to the plurality of input features associated with the second bot from among the at least two bots;
update a dynamic compatibility matrix with the at least two bots based on the end-user bot compatibility value and the technical compatibility value, wherein the dynamic compatibility matrix is updated with a value one if the end-user compatibility value and the technical feasibility value are set to one and zero otherwise; and
create a bot package by integrating the at least two bots based on the dynamic compatibility matrix, wherein the bot package is created only if a corresponding entry between the at least two bots in the dynamic compatibility matrix is set to one.
7 . The system of claim 6 , wherein the method of training the classifier for computing the end-user compatibility value of the at least two bots comprises:
receiving a plurality of bot features corresponding to each of a plurality of bots stored in the bot repository;
generating a preprocessed plurality of bot features corresponding to each of the plurality of bots using a preprocessing technique;
generating a plurality of vectorized bot features associated with each of the plurality of bots based on a corresponding preprocessed plurality of bot features using a vectorization technique;
generating a plurality of bot combinations based on the plurality of bots using a permutation combination technique; and
training the classifier based on the plurality of vectorized bot features associated with each of the plurality of bots and the plurality of bot combinations, wherein the classifier is trained to compute a compatibility score between each of the plurality of bot combinations.
8 . The system of claim 6 , wherein the plurality of bot features comprises a bot name, the plurality of input features, the plurality of output features, a datatype associated with each of the plurality of input features, a datatype associated with each of the plurality of output features and a bot category.
9 . The system of claim 6 , further comprises auto recommending a plurality of bots to the user using a Graph Neural Network (GNN) based self-learning model, wherein the GNN based self-learning model is associated with the dynamic compatibility matrix further comprising compatibility values of the plurality of bots.
10 . The system of claim 6 , wherein the created bot package is included in the bot repository.
11 . One or more non-transitory machine-readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors cause:
receiving at least two bots selected by a user from a bot repository as input, wherein each bot comprises a plurality of bot features;
computing an end-user bot compatibility value for the at least two bots based on a corresponding plurality of bot features using a trained classifier, wherein the end-user compatible value is set to one only if the corresponding end-user bot compatibility score is greater than a predefined threshold;
generating a technical compatibility value of the at least two bots based on a plurality of input features and a plurality of output features associated with the at least two bots, wherein the technical compatibility value is set to one only if the plurality of output features of a first bot from among the at least two bots is equal to the plurality of input features associated with the second bot from among the at least two bots;
updating a dynamic compatibility matrix with the at least two bots based on the end-user bot compatibility value and the technical compatibility value, wherein the dynamic compatibility matrix is updated with a value one if the end-user compatibility value and the technical feasibility value are set to one and zero otherwise; and
creating a bot package by integrating the at least two bots based on the dynamic compatibility matrix, wherein the bot package is created only if a corresponding entry between the at least two bots in the dynamic compatibility matrix is set to one.
12 . The one or more non-transitory machine-readable information storage mediums of claim 11 , wherein the method of training the classifier for computing the end-user compatibility value of the at least two bots comprises:
receiving a plurality of bot features corresponding to each of a plurality of bots stored in the bot repository;
generating a preprocessed plurality of bot features corresponding to each of the plurality of bots using a preprocessing technique;
generating a plurality of vectorized bot features associated with each of the plurality of bots based on a corresponding preprocessed plurality of bot features using a vectorization technique;
generating a plurality of bot combinations based on the plurality of bots using a permutation combination technique; and
training the classifier based on the plurality of vectorized bot features associated with each of the plurality of bots and the plurality of bot combinations, wherein the classifier is trained to compute a compatibility score between each of the plurality of bot combinations.
13 . The one or more non-transitory machine-readable information storage mediums of claim 11 , wherein the plurality of bot features comprises a bot name, the plurality of input features, the plurality of output features, a datatype associated with each of the plurality of input features, a datatype associated with each of the plurality of output features and a bot category.
14 . The one or more non-transitory machine-readable information storage mediums of claim 11 , wherein the one or more instructions which when executed by the one or more hardware processors further cause compatibility values of the plurality of bots.
15 . The one or more non-transitory machine-readable information storage mediums of claim 11 , wherein the created bot package is included in the bot repository.