ResolveX : AI-Powered Sentiment Analysis in Campus Grievance Systems

ResolveX : AI-Powered Sentiment Analysis in Campus Grievance Systems

Abhijna

Dept. of Computer Applications,

St. Joseph Engineering College, Mangaluru, India

Abstract – To preserve accountability and transparency in education, we must address issues. There are instances when traditional methods fail to convey the intended message with precision. Resolvex is an AI- based platform that leverages Re- act.js, FastAPI, and MongoDB to offer scalable security for users grievances in order manage them. The system uses Natural Language Processing (NLP) to analyze sentiment and order de- partments by complaint volume, resolution time frame, severity, and emotional tone. It enables the monitoring and submission of anonymous complaints. Additionally, predictive analytics can identify departments that are stagnant and performing poorly. Based on preliminary research on simulated datasets, it is concluded that ResolveX is a tactical tool with high accuracy, responsiveness, and usefulness for proactive institutional reform.

Index Terms – AI-Powered Systems, Complaint Management, Educational Transparency, FastAPI, Grievance Redressal, Mon- goDB, Natural Language Processing, Predictive Analytics, Re- act.js, Sentiment Analysis

1. INTRODUCTION

   A grievance redressal process is essential to educational institutions accountability, openness, and institutional responsiveness. 1. It offers students and staff a methodical way to voice concerns and seek solutions. For traditional grievance management methods, whether manual or based on basic digital technology, maintaining anonymity, promptly resolving grievances, and making fact-based decisions are significant obstacles.

   Traditional grievance procedures have a number of problems: The shortcomings of the grievance procedure

   • People are unsure of how or where to express their concerns.

   • The lack of privacy makes it more difficult for people to disclose sensitive information.

   • The status of user-submitted complaints cannot be tracked. The way complaints are handled has the following issues:

   • The automatic renewal grievances are not compiled or sorted.

   • The complaints content is vague and mislabeled.

   • A poor follow-up method for communicating or feedback.

   • Misdirected complaints are forwarded to the wrong de- partments.

   Murari B K,

   Dept. of Computer Applications,

   St. Joseph Engineering College, Mangaluru, India

   Resolvex, a grievance management platform that employs artificial intelligence, was created to address these concerns. For user interaction, the design is a frontend built around React.js, with backends powered by FastAPI and MongoDB for flexible data storage. A secure web interface is utilized to provide real-time tracking and anonymous submission of complaints.

   Natural Language Processing (NLP) employs techniques for sentiment analysis and classification as well. Metrics such as complaint volume, resolution time, severity, and emotional tone are used to rank departments. The use of predictive analytics enables the management team to identify areas of improvement, such as departments that are failing or reporting frequent problems.

   Resolvexs scope is intended for academic institutions, beginning with a multi-departmental prototype implementation. Using simulated data and early user input, we assess the systems accuracy, responsiveness, and usability.

2. LITERATURE REVIEW

   Many of the systems in use today are made to speed up the complaint management process, despite their limited automation, analytics, and intelligence capabilities.

   By utilizing the customary Software Development Life Cycle (SDLC) model, Sharnitha et al. suggested a Smart Complaint Management System.

   Kormpho et al. developed a Smart Complaint Management System (SCMS) that integrates chatbot interfaces with ma- chine learning-based routing . However, the system solely concentrated on maintenance-related complaints and lacked sentiment score and predictive insights.

   For college campuses, Patle et al. developed a dashboard- driven complaint system. Despite emphasising openness and usability, the system does not use artificial intelligence or natural language processing to identify reoccurring problems or evaluate user feedback.

   The proposed remedy, ResolveX, employs a modular, AI- integrated architecture to resolve these drawbacks. In addition to enabling anonymous reporting, it employs a rating algorithm based on the volume of complaints, the time it requires to resolve them, and the emotional tone of the complaints, as well as Natural Language Processing (NLP) for sentiment

   analysis. The systems capacity to identify persistent problems and under performing departments is further improved by predictive analytics. ResolveX is a sophisticated platform for institutional reform that offers proactive monitoring and updates in real time.

   TABLE I

   COMPARISON OF EXISTING SYSTEMS WITH THE PROPOSED RESOLVEX

   Paper & Year	Methodology Used	Features Implemented	Drawbacks
   Sharnitha et al. (2025)	Phased Waterfall SDLC with UAT	Report creation, role-based access, tracking, complaint filing	Limited analytics; lacks intelligent ranking, AI integration, and forecasts
   Pongpaiche et al. (2018)	Iterative prototyping with ML integration	Duplicate detection, automated routing, chatbot interface	No NLP emotion scoring; limited to maintenance; lacks ranking
   Patle et al. (2023)	User-centered SDLC with feedback loops	Complaint tracking, departmental dash- board, basic UI	No sentiment anal- ysis, AI alerts, or predictive features
   Proposed (Re- solvex)	Modular SDLC with AI/ML integration using agile- inspired iterations	Sentiment analysis, department ranking, anonymous complaints, predictive analytics	Scalable AI/ML- backed decision support

3. METHODOLOGY

   Resolvexs adaptable and scalable development process combines user-centered design with AI-powered analytics. The three main parts of the systemthe frontend interface, back- end API service, and AI/NLP analytics moduleare linked to a MongoDB database to provide flexible data storage and retrieval.

   1. System Architecture

      Resolvexs adaptable user interface, powered by a Re- act.js frontend, enables staff and students to file and monitor complaints. The backend, which was constructed using the fastAPI Python framework, manages data processing, routing, authentication, and communication with the AI services.The information is kept in MongoDB, a NoSQL database that offers scalability and schema variation.

      The architecture supports a wide range of user roles,including:

      • User: The user can report problems or log in anony- mously.

      • Department: Provides status updates and assesses com- plaints that are assigned to it.

      • Admin: Tracks departmental performance, maintains feedback information, and sends out alerts.

   2. Complaint Submission and Tracking

      Users can file grievances using an interactive online form that allows for anonymity ad provides options to select a department, indicate the sort of complaint, and include a detailed description. Following submission, each complaint is assigned a unique ID and stored in a database. Users can go

      Fig. 1. System architecture of ResolveX showing frontend, backend, database, AI module, and user roles.

      to a secure monitoring page in the system to see how their complaints are progressing.

   3. NLP-Based Sentiment Analysis

      ResolveX employs apre-trained deep knowledge model from the Hugging Face Mills library distilbert- base- uncased- finetuned- sst-2-english for sentiment analysis. The following is how the system operates:

      • However, if a complaint or piece of feedback is blank, it receives a neutral sentiment score of 0.0 is given.

      • However, if the model determines that the sentiment is positive, the score (which ranges from 0 to 1) is kept.

      • However, if the score is categorised as negative, it is negated (between 0 and -1).

        The following is the justification for implementation:

        Fig. 2. system that assigns a sentiment rating according to the model label and confidence output.

   4. Department Ranking and Analytics

      The departments performance is evaluated using a special ranking methodology that considers the following elements:

      • The number of complaints

      • The average time it takes to resolve them

      • The sentiment analysiss assessment of the complaints severity

      • The commenters emotional tone.

   These indicators help identify common complaint patterns and failing departments. They will be shown on the admin dashboard.

4. RESULTS AND DISCUSSION

   ResolveX is an AI-powered complaint management system that was designed, developed, and tested for a multi- departmental educational institution. Although it hasnt been used in a real campus setting yet, extensive simulation and system-level testing were conducted to evaluate its key com- ponents and expected outcomes.

   1. Expected Capabilities and Features

      The system allows anonymous users as well as registered users to file complaints. Complaints are automatically sorted and routed to the appropriate department based on predefined criteria. Each department can keep an eye on, handle, and fix issues thanks to the dashboard. After complaints have been resolved, users can leave comments, which the system assesses using sentiment analysis and natural language processing techniques.

      Through the admin interface, higher authorities can view rankings, track departmental performance, get alerts for issues that havent been fixed, and use AI-generated summaries to examine monthly patterns.

   2. Simulated Use Case and Test Results

      The ability to function was verified by running multiple test cases with simulated complaints. The system successfully completed the following tasks:

      • Handled grievances in a number of locations, such as the residence hall, office, and classroom.

      • Complaints that are automatically categorized and sent to the relevant departments.

      • Departments to update status, mark as resolved, and submit comments.

      • Department rankings each month using the sentiment scores from the feedback.

      • Displayed on the admin dashboard, along with a summary of the analytics and insights for every department.

   3. AI Module Performance

      The sentiment analysis model was tested using sample feedback data with known sentiment labels. In these test cases, the model was able to classify feedback into positive, neutral, and negative categories with reasonable accuracy ( 85%) based on labeled examples. Though trained on limited data, the model demonstrated the potential to support automated analysis of user satisfaction and department performance.

      To assess ResolveXs emotion classification skills, a confu- sion matrix was generated using a synthetic dataset of 241 labeled feedback entries. With an accuracy of 76. 76%, the DistilBERT- based classifier properly projected 102 negative and 83 positive emotions. Twelve false positives and forty- four false negatives existed. This analysis shows how well the model performs in sentiment-based feedback classifica- tion, hence supporting automated departmental performance monitoring.

   4. Estimated Advantages

      In a real-world setting, ResolveX ought to provide the following advantages:

      • Greater openness in the way complaints are handled

      • Problem resolution can be accelerated by forwarding issues straight to departments.

      • enhanced user participation as a result of the anonymous submission option.

      • Departmental performance appraisal based on data.

      • Less administrative effort for escalation and surveillance.

   Fig. 3. Confusion Matrix of Sentiment Classification (Accuracy = 76.76%).

5. CONCLUSION

A state-of-the-art, AI-powered approach to managing com- plaints in educational settings is ResolveX. By integrating technologies like React.js, FastAPI, MongoDB, and NLP- based sentiment analysis, the system addresses significant issues with current methods, such as limited performance in- sights, delayed resolutions, and lack of anonymity. Simulation testing confirmed its ability to categorize complaints, assess input, and generate department rankings based on significant characteristics. Although there is currently a paucity of prac- tical implementation, the technology has enormous potential to improve institutional responsiveness and decision-making. With future improvements like real-time notifications and language input support, ResolveX is well-positioned to grow into a helpful tool for proactive and data-driven complaint resolution.

REFERENCES

1. Sharnitha. P.M, Ajitha. I. SMART COMPLAINT MANAGEMENT SYS- TEM. Dr. N.G.P. Arts And Science College, Tamil Nadu, India. DOI: https://www.doi.org/10.56726/IRJMETS372804

2. Kormpho, Pattamaporn, et al. Smart Complaint Management Sys- tem.

   2018 Seventh ICT International Student Project Conference (ICT-ISPC), IEEE, 2018, pp. 16. DOI: https://doi.org/10.1109/ICT-

   ISPC.2018.8523949

3. Patle, Prof. Shekhar, et al. Efficient Campus Solutions: A Journal on Enhancing College Complaint Management. IJARCCE, vol. 12, no. 12,

   Dec. 2023. DOI: https://doi.org/10.17148/IJARCCE.2023.121213

4. Banga, Dishant, and Kiran Peddireddy. Artificial Intelligence for Cus- tomer Complaint Management. International Journal of Computer Trends and Technology, vol. 71, no. 3, Mar. 2023, pp. 16. DOI.org (Crossref), https://doi.org/10.14445/22312803/IJCTT-V71I3P101.

5. Ahmadi, Amirmasoud, et al. State-Based Decoding of Force Signals From Multi-Channel Local Field Potentials. IEEE Access, vol. 8, 2020, pp. 15908999. DOI.org (Crossref), https://doi.org/10.1109/ACCESS.2020.3019267.

6. Preface. Procedia Computer Science, vol. 172, 2020, pp. 12. DOI.org (Crossref), https://doi.org/10.1016/j.procs.2020.06.043.

7. Bendra Agustina, Elvinda, et al. Dependence of Microstructure and Optical Properties on Holding Time and Annealing Temperature of BiFeO3thin Film Fabricated by Chemical Solution Deposition (CSD). Materials Today: Proceedings, vol. 44, 2021, pp. 331318. DOI.org (Crossref), https://doi.org/10.1016/j.matpr.2020.11.534.

8. Mounika, Valasapalli, et al. Prediction of Type-2 Diabetes Using Ma- chine Learning Algorithms. 2021 International Conference on Artificial

   Intelligence and Smart Systems (ICAIS), IEEE, 2021, pp. 12731.

   DOI.org (Crossref), https://doi.org/10.1109/ICAIS50930.2021.9395985

9. Lieto, Antonio, et al. Integrating a Cognitive Framework for Knowledge epresentation and Categorization in Diverse Cognitive Architectures. Procedia Computer Science, vol. 71, 2015, pp. 9298. DOI.org (Crossref), https://doi.org/10.1016/j.procs.2015.12.211.

10. Chen, Junqiao. Model Algorithm Research Based on Python Fast API. Frontiers in Science and Engineering, vol. 3, no. 9, Sep. 2023, pp. 710. DOI.org (Crossref), https://doi.org/10.54691/fse.v3i9.5591.