Introduction
The Discussion section of a research paper is where authors get to interpret their findings, contextualize their research, and propose future directions. It is a crucial section that provides the reader with insights into the significance and implications of the study.
However, writing an effective Discussion section can be challenging, as it requires authors to not only summarize their results but also critically analyze them. In this blog post, we’ll provide you with 7 essential steps to writing an informative and thought-provoking Discussion section.
By following these steps, you’ll be able to write a compelling Discussion section that will not only enhance the reader’s understanding of your research but also contribute to the broader scientific community.
Focus on the Relevance
The purpose of every research is to implement the results for the positive development of the relevant subject. The discussion section should have these major factors listed before beginning to describe how the research was conceived and the sequence of developments that took place.
For example, a Robot navigation system has to handle a large amount of uncertain data in real-life environment. In the proposed experiment Fuzzy logic addresses this problem as it takes uncertain data, processes it and obtains certain and finite data.
Highlight the Limitations
If the research is on a subject that might have legal limitations or restrictions that might have caused certain imperfections, it should be acknowledged by the researcher before the work is criticized by others later in his/her discussion section.
Example: Object motion prediction is done by incorporating human experience in the form of fuzzy inference rules. It is assumed that the environment is observed through the stereo vision technique. The observed environment covers the semi-circular area in front of the Robot.
Introduce New Discoveries
Begin the discussion section by stating all the major findings in the course of the research. The first paragraph should have the findings mentioned, which is expected to be synoptic, naming and briefly describing the analysis of results.
Example: In the proposed work the predictor has been evaluated with three well-known defuzzification techniques. Based on the analysis of results, it has been found that the Mean Of Maximum defuzzification technique has a lower response time and better accuracy
Highlight the Observations
Coming to the major part of the findings, the discussion section should interpret the key observations, the analysis of charts and the analysis of tables. The researcher has to;
• Begin with explaining the objective of the research, followed by what inspired you as a researcher to study the subject.
• Explain the meaning of the findings, as every reader might not understand the analysis of graphs and charts as easily as people who are in the same field as you.
• The reader should be able to understand the key observations without being forced to go through the whole paper.
Case Study:
“An Optimized Fuzzy Based Short Term Object Motion Prediction for Real-Life Robot Navigation Environment” (Paper Link)
Discussion Section:
Figure 1 in the results section represent the average relative error observed for the prediction algorithm for various test cases using Min-Max, MOM and COA defuzzification techniques.

Figure 1: Short-term using Min-Max, MOM and COA defuzzification techniques.
For each test case, the average response time is also calculated to find its suitability to a
real-life environment(Figure 2).

Figure 2: Real-life video frames captured at INRIA lab
The prediction algorithm is tested by processing the real-life video frames (which are captured at every interval of 02 seconds). It is observed that the predictor with MOM defuzzification performs better in terms of response time and less relative error. The navigation environment is presented in the form of a prediction graph where the x-axis represents the range parameter and the y-axis represents the Angle parameter. The predicted Angle and Range values are compared with actual values obtained from real-life environments In most cases the predicted values are in the region of the actually measured range and Angle values.

Fig.3. Path Prediction using MoM Defuzzification Techniques
Figure 3. illustrates some of the results obtained for path prediction using MOM defuzzification. The performance of the predictor is tested when more than one object is sensed by the sensor. The response time of the predictor for all the objects should be acceptable for real-life applications. The tests are carried out assuming at most 6-8 objects can be visible and can affect the decisions to be made regarding robot traversal(Figure 4).

Figure 4: Comparison of Defuzzification techniques
In real-life applications, the Robot has to deal with multiple numbers of objects and the total response time for all the objects’ motion predictions should be less than the time gap between two sensor readings. The proposed predictor is able to generate results suitable for real-life situations.
Compare and Relate with other Research Works
No one is ever the only person researching a particular subject. A researcher always has companions and competitors. The discussion section should have a detailed comparison of the research. It should present the facts that relate the research to studies done on the same subject.
Example: The table below compares some of the well-known prediction techniques with our fuzzy predictor with MOM defuzzification for response time, relative error and Environmental constraints. Based on the results obtained it can be concluded that the Fuzzy predictor with MOM defuzzification has a less relative error and quick response time as compared to other prediction techniques. The proposed predictor is more flexible, simple to implement and deals with noisy and uncertain data from real-life situations. The relative error of 5-10% is acceptable for our system as the predicted fuzzy region and the fuzzy region of the actual position remains the same.
Table 1: Comparison of well-known Robot Motion prediction Techniques
Short Term Predictor | Environmental constraints if any | Relative Error | Response time in seconds |
ANN Predictor | A simulated environment with Rectilinear paths | 6-17% | — |
Bayesian Occupancy filter | Only for small-scale environments | Not specified | 100 x 10-3 |
Polynomial NN | Simulated environment | 1-10% | Not specified |
Auto Regressive model | Simulated Environment | Not specified | Computationally intensive |
Fuzzy Predictor with MOM | Real-life environment | 1-10% | 07×10-3 sec to 09×10-3 sec |
Provide Alternate View Points
Almost every time, it has been noticed that analysis of charts and graphs shows results that tend to have more than one explanation. The researcher must consider every possible explanation and potential enhancement of the study from alternative viewpoints. It is critically important that this is clearly put out to the readers in the discussion section.
Example: On critical observation of the graphs in environments where the number of objects is less and response time is critical Min-Max method can be used as its response time is better as compared to COA and MOM methods.
Show Future Directions
The section must have suggestions for research that should be done to unanswered questions. These should be suggested at the beginning of the discussion section to avoid questions being asked by critics. Emphasizing the importance of following future directions can lead to new research as well.
Example: The authors are in the process of further optimizing the rule base and improving the response time of the predictor. Further improvements in relative error parameter optimization of Min-Max method may lead to a highly efficient MinMax Predictor.
Phrases that can be used in the Discussion Section of a Research Paper
Here are some common phrases that can be used in the discussion section of a paper or research article. I’ve included a table with examples to illustrate how these phrases might be used:
Phrase | Example |
---|---|
Interpretation: This phrase is used to explain the meaning and significance of the results. | “The results suggest that increasing the number of hidden layers in a neural network can lead to higher accuracy on certain types of datasets, but may lead to overfitting on others.” |
Comparison to previous research: This phrase is used to compare the results to previous research in the field. | “Our findings are consistent with previous research on the effectiveness of ensemble methods for classification tasks (Smith et al., 2019; Jones et al., 2020).” |
Limitations: This phrase is used to describe limitations of the study or potential sources of error or bias. | “One limitation of our study is that we only evaluated the models on a single dataset, which may not generalize to other domains or applications.” |
Implications: This phrase is used to discuss the practical or theoretical implications of the results. | “The findings of this study could inform the development of more accurate and reliable systems for detecting fraud in financial transactions.” |
Future work: This phrase is used to suggest directions for future research or improvements to the current system or approach. | “Future work could explore the use of more complex feature engineering techniques to improve the performance of the machine learning models on imbalanced datasets.” |
Contributions: This phrase is used to describe the original contributions of the study or the novelty of the approach or methodology. | “To the best of our knowledge, this is the first study to evaluate the performance of a hybrid approach combining deep learning and reinforcement learning for autonomous driving in complex environments.” |
Strengths: This phrase is used to highlight the strengths or advantages of the current approach or methodology. | “One of the strengths of our approach is its ability to handle noisy and incomplete data, which is common in real-world applications.” |
Phrases that can be used in the Analysis part of the Discussion Section of a Research Paper
Here are some common academic phrases that can be used in the analysis section of a paper or research article. I have included a table with examples to illustrate how these phrases might be used:
Phrase | Example |
---|---|
Descriptive statistics: This phrase is used to describe the basic characteristics of the data, such as mean, standard deviation, and range. | “The average processing time for the proposed algorithm was 3.2 seconds, with a standard deviation of 0.5 seconds.” |
Inferential statistics: This phrase is used to describe the statistical tests used to draw conclusions from the data. | “We used a two-tailed t-test to compare the performance of the two algorithms, with a significance level of 0.05.” |
Correlation analysis: This phrase is used to describe the relationships between variables in the data. | “We found a strong positive correlation between the number of training samples and the accuracy of the classification model.” |
Regression analysis: This phrase is used to describe the relationships between one or more independent variables and a dependent variable. | “We used a multiple linear regression model to predict the processing time of the algorithm based on the number of input parameters and the complexity of the data.” |
Classification analysis: This phrase is used to describe the process of assigning observations to predefined categories. | “We evaluated the performance of the classification model using metrics such as precision, recall, and F1-score.” |
Clustering analysis: This phrase is used to describe the process of grouping similar observations together based on their characteristics. | “We used a k-means clustering algorithm to group the customers into four distinct segments based on their purchasing behavior.” |
Visualization: This phrase is used to describe the use of graphs or charts to illustrate patterns or relationships in the data. | “The scatter plot showed a clear positive correlation between the size of the training set and the accuracy of the classification model.” |
Conclusion
The Discussion section of a research paper is an essential part of any study, as it allows the author to interpret their results and contextualize their findings. To write an effective Discussion section, authors should focus on the relevance of their research, highlight the limitations, introduce new discoveries, highlight their observations, compare and relate their findings to other research works, provide alternate viewpoints, and show future directions.
By following these 7 steps, authors can ensure that their Discussion section is comprehensive, informative, and thought-provoking. A well-written Discussion section not only helps the author interpret their results but also provides insights into the implications and applications of their research.
In conclusion, the Discussion section is an integral part of any research paper, and by following these 7 steps, authors can write a compelling and informative discussion section that contributes to the broader scientific community.