## Bar Chart: Hits@1 Comparison of GoG and Graph-RFT at Different IKG Percentages ### Overview The image is a bar chart comparing the "Hits@1" metric for two methods, "GoG" and "Graph-RFT", across three different "IKG" percentages: 20%, 40%, and 60%. The chart uses vertical bars to represent the "Hits@1" values, with different colors distinguishing the two methods. ### Components/Axes * **Y-axis:** Labeled "Hits@1", with a numerical scale ranging from 0 to 80 in increments of 20. * **X-axis:** Categorical axis representing the "IKG" percentages: "IKG-20%", "IKG-40%", and "IKG-60%". * **Legend:** Located in the top-right corner, indicating that light green bars represent "GoG" and light blue bars represent "Graph-RFT". * **Gridlines:** Horizontal gridlines are present at intervals of 20 on the Y-axis. ### Detailed Analysis Here's a breakdown of the "Hits@1" values for each method at each "IKG" percentage: * **IKG-20%:** * GoG (light green): Approximately 45 * Graph-RFT (light blue): Approximately 72 * **IKG-40%:** * GoG (light green): Approximately 44 * Graph-RFT (light blue): Approximately 67 * **IKG-60%:** * GoG (light green): Approximately 36 * Graph-RFT (light blue): Approximately 60 **Trends:** * **GoG:** The "Hits@1" value decreases as the "IKG" percentage increases. * **Graph-RFT:** The "Hits@1" value decreases as the "IKG" percentage increases. ### Key Observations * Graph-RFT consistently outperforms GoG across all IKG percentages. * The difference in "Hits@1" between Graph-RFT and GoG is largest at IKG-20% and smallest at IKG-60%. * Both methods show a decrease in "Hits@1" as the "IKG" percentage increases. ### Interpretation The data suggests that Graph-RFT is a more effective method than GoG for the task being evaluated, as indicated by the higher "Hits@1" values across all "IKG" percentages. The decrease in "Hits@1" for both methods as the "IKG" percentage increases could indicate that higher "IKG" percentages introduce more noise or complexity that negatively impacts the performance of both methods, but affects GoG more significantly. The "IKG" percentage likely represents a parameter or condition within the experiment that influences the performance of the methods. Further context on what "IKG" represents would be needed for a more complete interpretation.

\n ## Bar Chart: Hits@1 Performance Comparison ### Overview This bar chart compares the Hits@1 performance of two models, GoG and Graph-RFT, across three different IKGs (IKG-20%, IKG-40%, and IKG-60%). The chart uses bar groupings to visually represent the performance of each model for each IKG. ### Components/Axes * **Y-axis:** "Hits@1" - Represents the performance metric, ranging from 0 to 80. * **X-axis:** IKG levels: "IKG-20%", "IKG-40%", "IKG-60%" * **Legend:** Located in the top-right corner. * Yellow: "GoG" * Light Green: "Graph-RFT" * **Gridlines:** Horizontal gridlines are present to aid in reading the values. ### Detailed Analysis The chart consists of six bars, grouped by IKG level. * **IKG-20%:** * GoG (Yellow): Approximately 48. The bar reaches slightly above the 40 mark, but is below the 50 mark. * Graph-RFT (Light Green): Approximately 72. The bar reaches slightly below the 75 mark. * **IKG-40%:** * GoG (Yellow): Approximately 42. The bar reaches slightly above the 40 mark, but is below the 45 mark. * Graph-RFT (Light Green): Approximately 68. The bar reaches slightly below the 70 mark. * **IKG-60%:** * GoG (Yellow): Approximately 32. The bar reaches slightly above the 30 mark, but is below the 35 mark. * Graph-RFT (Light Green): Approximately 64. The bar reaches slightly below the 65 mark. ### Key Observations * Graph-RFT consistently outperforms GoG across all three IKG levels. * The performance of both models appears to decrease as the IKG percentage increases. The largest drop in performance is observed between IKG-40% and IKG-60% for both models. * The difference in performance between the two models is most pronounced at the IKG-20% level. ### Interpretation The data suggests that Graph-RFT is a more effective model than GoG for this particular task, as measured by Hits@1. The decreasing performance of both models as the IKG percentage increases could indicate that the task becomes more challenging with larger or more complex knowledge graphs. The initial high performance at IKG-20% and subsequent decline could be due to the models overfitting to the smaller knowledge graph, or that the added complexity of the larger graphs introduces noise or irrelevant information. The consistent outperformance of Graph-RFT suggests it may be more robust to these challenges. Further investigation would be needed to understand the specific reasons for these trends and to determine whether the observed differences are statistically significant.

## Grouped Bar Chart: Performance Comparison of GoG and Graph-RFT Across IKG Percentages ### Overview The image displays a grouped bar chart comparing the performance of two methods, labeled "GoG" and "Graph-RFT," across three different conditions or datasets. The performance metric is "Hits@1," plotted on the y-axis. The chart visually demonstrates that Graph-RFT consistently achieves a higher Hits@1 score than GoG in all three categories, though both methods show a declining trend as the category percentage increases. ### Components/Axes * **Chart Type:** Grouped vertical bar chart. * **Y-Axis:** * **Label:** "Hits@1" * **Scale:** Linear scale from 0 to 80, with major tick marks at intervals of 20 (0, 20, 40, 60, 80). * **X-Axis:** * **Categories:** Three distinct categories are labeled: 1. `IKG-20%` 2. `IKG-40%` 3. `IKG-60%` * **Legend:** * **Position:** Top-right corner of the chart area, enclosed in a box. * **Entries:** 1. **GoG:** Represented by a light green (pale lime) color. 2. **Graph-RFT:** Represented by a teal (blue-green) color. * **Grid:** A light gray grid is present, with horizontal lines aligned to the y-axis ticks. ### Detailed Analysis The chart presents paired bars for each x-axis category. The left bar in each pair is GoG (light green), and the right bar is Graph-RFT (teal). **1. Category: IKG-20%** * **GoG (Light Green):** The bar height is approximately **45**. (Visual estimate: slightly above the 40 line). * **Graph-RFT (Teal):** The bar height is approximately **72**. (Visual estimate: well above the 60 line, closer to 72 than 70 or 75). * **Trend:** Graph-RFT significantly outperforms GoG in this category. **2. Category: IKG-40%** * **GoG (Light Green):** The bar height is approximately **44**. (Visual estimate: appears very slightly lower than the GoG bar for IKG-20%). * **Graph-RFT (Teal):** The bar height is approximately **67**. (Visual estimate: above the 60 line, but lower than its value for IKG-20%). * **Trend:** Both methods show a slight decrease compared to IKG-20%. The performance gap remains large. **3. Category: IKG-60%** * **GoG (Light Green):** The bar height is approximately **36**. (Visual estimate: clearly below the 40 line). * **Graph-RFT (Teal):** The bar height is approximately **60**. (Visual estimate: aligns almost exactly with the 60 grid line). * **Trend:** Both methods show their lowest performance in this category. The decline from IKG-40% to IKG-60% is more pronounced for Graph-RFT than for GoG. ### Key Observations 1. **Consistent Superiority:** Graph-RFT achieves a higher Hits@1 score than GoG in all three measured conditions (IKG-20%, IKG-40%, IKG-60%). 2. **Performance Degradation:** Both methods exhibit a downward trend in performance as the IKG percentage increases from 20% to 60%. 3. **Magnitude of Gap:** The absolute performance gap between Graph-RFT and GoG is largest at IKG-20% (~27 points) and narrows slightly at IKG-60% (~24 points), primarily because Graph-RFT's score drops more steeply. 4. **Relative Stability of GoG:** The GoG method's performance declines more gradually (from ~45 to ~36) compared to Graph-RFT's steeper decline (from ~72 to ~60). ### Interpretation The data suggests that the "Graph-RFT" method is substantially more effective than the "GoG" method for the task measured by the "Hits@1" metric across the tested scenarios. The "IKG" variable (likely representing some form of "Incomplete Knowledge Graph" or similar concept with a percentage of missing or corrupted data) has a negative impact on both methods' performance. The consistent decline indicates that as the IKG condition becomes more severe (higher percentage), the task becomes more difficult for both models. The fact that Graph-RFT maintains a significant lead even as performance degrades implies it is more robust to the challenges introduced by the IKG conditions. However, its steeper absolute decline might suggest its performance is more sensitive to the *degree* of IKG compared to GoG, which shows a flatter, more stable, but lower-performing response curve. This chart would be critical in a technical document to argue for the adoption of Graph-RFT over GoG, while also noting the shared vulnerability to increasing IKG percentages.

## Bar Chart: Hits@1 Comparison Between GoG and Graph-RFT Across IKG Thresholds ### Overview The chart compares the performance of two methods, **GoG** (light green bars) and **Graph-RFT** (teal bars), in terms of **Hits@1** metric across three IKG (Influence Knowledge Graph) thresholds: 20%, 40%, and 60%. The y-axis represents Hits@1 scores (0–80), while the x-axis categorizes data by IKG percentages. --- ### Components/Axes - **Legend**: Located in the top-right corner, with: - **GoG**: Light green (#e6f0d9) - **Graph-RFT**: Teal (#a1de93) - **X-axis**: Labeled "IKG-20%", "IKG-40%", "IKG-60%" (left to right). - **Y-axis**: Labeled "Hits@1", scaled from 0 to 80 in increments of 20. - **Bars**: Paired bars for each IKG category, with GoG on the left and Graph-RFT on the right. --- ### Detailed Analysis 1. **IKG-20%**: - **GoG**: ~45 Hits@1 (light green bar height aligns with ~45 on y-axis). - **Graph-RFT**: ~72 Hits@1 (teal bar height aligns with ~72 on y-axis). 2. **IKG-40%**: - **GoG**: ~44 Hits@1 (slightly shorter than IKG-20%). - **Graph-RFT**: ~67 Hits@1 (shorter than IKG-20% but still dominant). 3. **IKG-60%**: - **GoG**: ~36 Hits@1 (notable drop from previous categories). - **Graph-RFT**: ~60 Hits@1 (smallest drop among all categories). --- ### Key Observations - **Graph-RFT consistently outperforms GoG** across all IKG thresholds, with a performance gap of ~27 Hits@1 at IKG-20% narrowing to ~24 Hits@1 at IKG-60%. - **GoG shows a declining trend** as IKG increases (45 → 44 → 36), while **Graph-RFT declines more gradually** (72 → 67 → 60). - The largest absolute difference occurs at IKG-20% (27 Hits@1), but the relative gap narrows slightly at higher IKG thresholds. --- ### Interpretation The data suggests that **Graph-RFT is more robust or efficient** than GoG in handling IKG thresholds, particularly at higher IKG levels (60%). The narrowing performance gap at IKG-60% implies that Graph-RFT may scale better with increased complexity or noise in the IKG. GoG’s sharper decline at IKG-60% could indicate sensitivity to higher IKG densities or structural challenges in the graph. These trends highlight the importance of method selection based on IKG characteristics in applications like recommendation systems or knowledge graph analysis.