This repository contains the code and data for a study that aimed to cluster sentences based on semantic similarities using FMRI (functional magnetic resonance imaging) data. The study initially attempted to use supervised learning methods but did not achieve satisfactory results. As a result, unsupervised learning techniques were employed to uncover intriguing findings related to meaningful patterns associated with the definitions and usages of objects within the sentence clusters.
The objective of this study was to gain insights into how the brain processes information by analyzing FMRI data. The researchers hypothesized that clustering sentences based on semantic similarities could provide valuable insights into information processing mechanisms. The study focused on both the meanings and usages of objects within the sentences.
Data Collection: FMRI data was collected from a group of participants while they were presented with various sentences.
Data Preprocessing: The FMRI data underwent preprocessing steps to ensure data quality and to extract relevant features for analysis.
Supervised Learning Attempts: Initially, supervised learning methods were applied to the data in an attempt to cluster sentences based on semantic similarities. However, the results did not meet the desired criteria.
Unsupervised Learning Techniques: To overcome the limitations of supervised learning, unsupervised learning techniques were employed. These techniques revealed intriguing patterns and clusters related to the definitions and usages of objects within the sentences.
The findings of this study suggest that unsupervised learning methods can be more effective than supervised learning in clustering sentences based on semantic similarities. The discovered patterns and clusters shed light on how the brain processes information related to the meanings and usages of objects within sentences.
While this study has provided valuable insights, further research is necessary to enhance our understanding of information processing mechanisms. Future directions include:
Incorporating data from multiple participants to improve the generalizability of the findings. Developing enhanced frameworks and algorithms to further explore the patterns and clusters observed in the FMRI data. Conducting additional experiments to validate and refine the results obtained in this study. The code and data in this repository can serve as a foundation for future studies aiming to investigate the semantic processing of sentences using FMRI data.
data/: This directory contains the FMRI data collected during the study. Due to privacy concerns, the data has been anonymized and stripped of any personally identifiable information. code/: This directory contains the code used for data preprocessing, supervised learning attempts, and unsupervised learning techniques. The code is provided in Python and is well-documented to facilitate understanding and reproducibility. Conclusion This study has demonstrated the potential of unsupervised learning techniques in clustering sentences based on semantic similarities using FMRI data. The findings contribute to our understanding of how the brain processes information and highlight the need for further research to advance our comprehension of information processing mechanisms. Researchers and practitioners in the field can leverage the code and data provided in this repository for future investigations in this domain.