Publications

1. Improvising Language Conversation – Hindi & Urdu

In a diverse country such as India, natural language processing (NLP) plays an even greater role in resolving the complexities of languages. In this article, we discuss Urdu terminology used by Hindi literature and explain the problem for readers that are not familiar with these phrases. We devised an automated manner to replace Urdu based words in Hindi sentences with synonymous terms used commonly in conversational or literary Hindi. This study supports current attempts at standardization and simplification that have been underway for Hindi, encouraging linguistic diversification in sectors like the media or education. While there are challenges to solve for, such as handling homonyms and a fixed list of synonyms available, we empirically show strong improvements towards more inclusive language. Overall, the research presents a starting point for future multilingual application based work and highlights that NLP holds exciting promise to support reliable communication processes within diverse linguistic domains.

Reference: S. Agarwal, L. A. R. M and S. Dutta, "Improvising Language Conversation – Hindi & Urdu," 2025 IEEE International Students' Conference on Electrical, Electronics and Computer Science (SCEECS), Bhopal, India, 2025, pp. 1-5,

2. Comparative Study of NLP Tools for the Detection of Cyber Bullying

Social networking sites have increased in popularity in recent years because they provide users with a platform where they can connect worldwide and discuss their interests. However, people will benefit from a good environment thanks to the social network. Random cyberbullying is wild, and it poses a serious threat to the physical and mental health of the victim. It is important to find the right protection to detect and prevent this. The proposed method aims to detect the presence of cyberbullying activities on social networks such as Twitter where the tweets are scrapped and made it into a clean dataset and preprocessed the data by removing the “stopwords,” “lemmatization,” “tokenization” and more which is explained further. The dataset is then compared using the “NLP models” (e.g., ‘stanza’, ‘Roberta’, ‘NLTK’). In this article, we also compare the performance of three different NLP tools in detecting cyberbullying on Twitter profiles. We evaluated the tool using data from recorded tweets. We found that NLTK outperforms “Stanza” and “RoBERTa” in detecting cyberbullying. “NLTK” achieved an accuracy of 94%, while “Stanza” and “RoBERTa” together achieved an accuracy of 19%.

Reference: Bhadra, J., Kashyap, N., Poovamma, K., & Dutta, S. (2024, Feb). Comparative Study of NLP Tools for the Detection of Cyber Bullying. In International Conference on Emerging Research in Computation, Information, Communication, AI and ML ( 311-318). Springer Nature

3. Network under Deep Uncertainty concerning Food Security

We propose new models to find the vulnerable countries in terms of food security. These models are based on network analysis under deep uncertainty. The conditions of deep uncertainty affect the supply and demand of food, namely, carbohydrates of the countries. Under such conditions networks of the carbohydrate supply between countries are constructed. The countries vulnerability in terms of food security were studied by new centrality indices taking into account carbohydrate consumptions of countries and the possibility of group influence of countries to a country. Also, our models show direct and indirect dependence on import of carbohydrates from other countries. The scenario of one of such situations was constructed, and our models for studying this situation were tested. The vulnerable countries are identified in terms of carbohydrate consumption from main crops in different scenarios based on real data using our new models. The developed models can make the food policy of countries more efficient.

Reference: F. T Aleskerov. et.al , Journal of the New Economic Association, Vol- 64, No-3, 2024 page - 12-29

4.  Multi-Objective Fuzzy Probabilistic Programming Approach for Obtaining Crops Pattern with Water Replenishment

This research paper is concerned with the solution procedure of a multi-objective fuzzy probabilistic problem (MOFPP), taking different crop patterns as the objective functions and water requirements of crops as constraints, which is then solved using fuzzy programming technique. In order to handle fuzzy probabilistic constraints, a transformation technique is developed for fuzzy uniform distribution. As the study area receives less rainfall, the cultivation is mainly dependent on irrigation water. A continual use of underground water for irrigation led to a substantial decrease in the water level of the surrounding area. So, to uplift the underground water level, a proper selection of crops can play a vital role in the replenishment of underground water without compensating the farmer’s income. To illustrate the above technique, a case study is provided with the cultivated area of different crops as a decision variable. The two objective functions are taken as the yield of crops taken as fuzzy number and profit obtained by the farmers after selling the products. A comparison table is furnished in the result section for making better decisions. The study reveals that planting a combination of beans, bitter-gourd, and cabbages will yield maximum profit.

Reference: Dutta, S., Sahoo, B. C., Bhanavi, S., & Nethra, S. (2024). Multi-Objective Fuzzy Probabilistic Programming Approach for Obtaining Optimum Crops Pattern with Water Replenishment. International Jr of Applied and Computational Mathematics, 10(4), 128.

5.  A Novel Approach to Solve Multi‐objective Fuzzy Stochastic Bilevel Programming Using Genetic Algorithm

A bilevel programming is a two-level optimization problem, namely, the upper level (leaders) and the lower level (followers). The two level’s decision variables are entwined with each other which increases the complexity to obtain the global solution for both the optimization problems. Each level aims to optimize their own objective function under the given constraints at both the levels. To reduce the complexity partial cooperation between the two levels has been exploited in obtaining the Pareto solution. A novel solution procedure is proposed for a multi-objective fuzzy stochastic bilevel programming (MOFSBLP) problem is studied and solved using genetic algorithm. In this paper, previous information of the lower level is used as a fuzzy stochastic constraints in the upper level along with its constraints. Then with the solution of the combine constraints, the lower level solution is evaluated. The proposed solution procedure is illustrated by a numerical example taken from Zheng et al., and results are compared. A simpler version is solved using GAMs software to analyze the result of the numerical example. The proposed method highlights the importance of partial cooperation in solving bilevel programming problem. The advantage of the proposed solution method is that it creates common constraint space which helps in convergence of the algorithm.

Reference: Dutta, S., & Acharya, S. (2024, February). A Novel Approach to Solve Multiobjective Fuzzy Stochastic Bilevel Programming Using Genetic Algorithm. In Operations Research Forum (Vol. 5, No. 1, p. 11). Cham: Springer International Publishing.

6. Network under Deep Uncertainty

We construct a model to find the insecure countries in terms of food security using network analysis under the conditions of deep uncertainty. The conditions under deep uncertainty can be drought, flood, earthquakes, etc., which affect the supply and demand of food of the countries. Under such conditions which countries are more vulnerable has been studied. In order to study such situations, a model for healthy consumption of different products has been built, and a network is constructed to determine the vulnerable countries using the import and export data of basic food products (Rice, Wheat, Maize, Sorghum, Barley and Rye) for various countries. Different measures of network analysis have been applied and studied to discover the dependency of one country from other country or countries for food supply. The result of the study will definitely be helpful to the countries to resist such threats beforehand.

Reference: Aleskerov, F., Dutta, S., Egorov, D., & Tkachev, D. (2022). Networks under deep uncertainty. Procedia Computer Science, 214, 1285-1292.

7. Cubic Spline Interpolation Approach to Solve Multi‐ Choice Programming Problem

Multi-choice has become a significant part of the real-life decision-making process. Most of the problems involve more than one parameter as a choice, and among those different choices only one choice is to be made, which will optimize the objective function. The difficulty in making such a choice can be at ease with the help of mathematical techniques. In this paper, we propose a novel solution procedure to handle the multi-choice parameters in the constraint using cubic spline interpolation method. After analyzing the results, we observed that the proposed method yields better results as compared to existing methods. Two numerical examples are presented to explain the method and validate the fact of complete utilization of the resources.

Reference: Dutta, S., & Kaur, A. (2023). Cubic Spline Interpolation Approach to Solve Multi-Choice Programming Problem. International Journal of Applied and Computational Mathematics, 9(1), 6.

8. Multi‐Choice Programming with Benefits using Kriging Method

Most of the real-life problems involves multi-choice or more than one option, and in such a situation out of many choices, one choice needs to be selected for making decision. In this paper, we propose a novel problem involving multiple choice parameter with benefit and to handle such multiple-choice parameter, Kriging interpolation method is used, as other existing mathematical method cannot be used. An illustration is provided to explain the proposed method. Proc of the International Conference on Computational Intelligence and Sustainable Technologies.

Reference: Dutta, S., & Nair, R. V. (2022, February). Multi-choice programming with benefits using kriging interpolation method. In Proc of the International Conference on Computational Intelligence and Sustainable Technologies: ICoCIST (1-10). Springer Nature.

9. Fuzzy Programming Approach to a Multi‐Objective Fuzzy Stochastic Routing and Siting Hazardous Wastes

The aim of the research article is not only to propose a solution procedure to solve multi-objective fuzzy stochastic programming problem by using genetic-algorithm-based fuzzy programming method, but also to apply the computational techniques for transportation of the hazardous waste materials. In this article, routing and siting problems for nuclear hazardous waste material are studied and solved. The amount of waste materials generated in the nuclear reactors follows normal distribution. The two considered objective functions are about route selection which includes minimum travel time and minimum number of houses along the way, taking the safety measures into consideration. A multi-objective fuzzy stochastic mathematical model is formulated with the above mentioned objective functions and the route selection as the constraints. The proposed solution procedure is illustrated by a numerical example and a case study.

Reference: Dutta, S., Acharya, S., & Mishra, R. (2020). Fuzzy programming approach to a multi-objective fuzzy stochastic routing and siting hazardous wastes. Transportation Management, 3(1), 1-24.

10. Fuzzy stochastic price scenario based portfolio selection and its application to BSE using genetic algorithm

This paper is concerned with portfolio selection problem using a fuzzy stochastic price scenario. In this scenario, a ratio factor (k) is calculated from the historical data to generated the future price of the stocks of Bombay Stock Exchange. The ratio factor k of different stocks are treated as a fuzzy numbers, which in turn gives future fuzzy prices of the stocks. Returns on the stocks are calculated from the future price of the stocks. Rejection of the assets are done based on returns calculated from the worst case scenario. If the returns of an asset exceed the investor's risk tolerance then the asset are not included in the portfolio. The definition of capital budget has been reformed to include the transaction cost with the capital budget. This process is implemented in two stage multi-objective fuzzy probabilistic programming problem which is then solved using a fuzzy genetic algorithm to obtain maximum short term and long term returns. A case study of Bombay Stock Exchange is provided to illustrate the above model.

Reference: Dutta, S., Biswal, M. P., Acharya, S., & Mishra, R. (2018). Fuzzy stochastic price scenario based portfolio selection and its application to BSE using genetic algorithm. Applied Soft Computing, 62, 867-891.

11. Fuzzy Stochastic Genetic Algorithm for Obtaining Optimum Crops Pattern and Water Balance in a Farm

This paper is concerned with multi-objective fuzzy stochastic model for determination of optimum cropping patterns with water balance for the next crop season. The objective functions of the model is to study the effect of various cropping patterns on crop production subject to total water supply in a small farm. The decision variables are the cultivated area of different crops at the farm. The water requirement of the crops follows fuzzy uniform distribution and yields in the objective functions are taken as a fuzzy numbers. The model is solved by using fuzzy stochastic simulation based genetic algorithm without deriving the deterministic equivalents.

Reference: Dutta, S., Sahoo, B. C., Mishra, R., & Acharya, S. (2016). Fuzzy stochastic genetic algorithm for obtaining optimum crops pattern and water balance in a farm. Water Resources Management, 30(12), 4097-4123.

12. Genetic algorithm based fuzzy stochastic transportation programming problem with continuous random variables

This paper is concerned with the solution procedure of a multi-objective transportation problem with fuzzy stochastic simulation based genetic algorithm. Supplies and demands are considered as a fuzzy random variables with fuzzy means and fuzzy variances in proposed multi-objective fuzzy stochastic transportation problem. The first step in fuzzy simulation based genetic algorithm is to deal with aspiration level of the constraints with the help of alpha-cut technique to obtain multi-objective stochastic transportation problem. In next step, fuzzy probabilistic constraints (fuzzy chance constraints) are handled within fuzzy stochastic simulation based genetic algorithm to obtain a feasible region. The feasibilities of the chance constraints are checked by the stochastic programming with the genetic process without deriving the deterministic equivalents. The feasibility condition for the transportation problem is maintained through out the problem. Finally, multiple objective functions are considered in order to generate a Pareto optimal solutions for the fuzzy stochastic transportation problem using the proposed algorithm. The proposed procedure is illustrated by two numerical examples.

Reference: Dutta, S., Acharya, S., & Mishra, R. (2016). Genetic algorithm based fuzzy stochastic transportation programming problem with continuous random variables. Opsearch, 53(4), 835-872.

13. Genetic algorithm approach for solving multi‐objective fuzzy stochastic programming problem

This paper is concerned with the solution procedure of a multi-objective fuzzy stochastic optimization problem by simulation-based genetic algorithm. In this article, a multi-objective fuzzy chance constrained programming problem is considered with continuous fuzzy random variables. The uncertain parameters are considered as fuzzy normal and fuzzy log-normal random variables. The feasibilities of the fuzzy chance constraints are checked by the fuzzy stochastic programming with the genetic process without deriving the deterministic equivalents. The proposed procedure is illustrated by a numerical example.

Reference: Dutta, S., Acharya, S., & Mishra, R. (2017). Genetic algorithm approach for solving multi-objective fuzzy stochastic programming problem. International Journal of Mathematics in Operational Research, 11(1), 1-28.

14. Computation of a Multi‐ Choice Fuzzy Goal Programming Problem

The important activity of a manager is Decision Making. Decision Making is getting complex day by day due to incomplete information and conflicting criteria. Multi Choice Goal Programming (MCGP) is considered as an important tool in multi criteria decision making to solve such problems. However in real world problems, finding accurate targets for a goal is a difficult task. In order to deal with this type of uncertainty, we consider multi-choice fuzzy target values. The concept of multiple number of fuzzy target values lead to multi-choice fuzzy goal programming (MCFGP) problem. MCFGP cannot be solved by existing methods. In order to solve MCFGP, its equivalent mathematical model is presented by using three different techniques. Finally the equivalent models become mixed integer non linear mathematical models. In order to solve the mixed integer non linear programming model, the help of existing optimization software have been taken. To illustrate the methodology a numerical example is provided.

Reference: Kannan Kumar Patro, Mitali Acharya, Sanjay Dutta, S. Acharya, Computation of a Multi‐ Choice Fuzzy Goal Programming Problem, Global journal of pure and applied mathematics, Vol‐ 11, No. 6, 2015, pp. 4207‐4227.