Question about Table 2 in the paper!

[cskkxjk]

HI, thank you for your work! I was wondering if you could provide the plot code for table 2 in the paper.
I'm a bit confused by the phrase 'the radius reflecting the performance of each system' in the table.

[wuchuanjie]

Thank you for your interest in our work! The experiment presented in Figure 2 mainly compares Vanilla RAG with recent GraphRAG systems (MS-GraphRAG, HippoRAG, and GFM-RAG) on the G-Medical dataset. Each circle corresponds to the performance of one RAG system. The position of a circle’s center represents its retrieval performance, while the radius of the circle indicates the QA accuracy.

plot code：

import pandas as pd
import matplotlib.pyplot as plt

df = pd.DataFrame(
    {
        "method": ["Vanilla RAG", "MS-GraphRAG", "HippoRAG", "GFM-RAG", "LightRAG"],
        "llm_acc": [61.07, 55.67, 57.06, 57.06, 56.42],
        "recall": [71.85, 78.76, 84.04, 84.30, 82.57],
        "relevance": [62.87, 28.68, 47.19, 38.50, 43.08],
    }
)

SHOW_LIGHTRAG = False
if not SHOW_LIGHTRAG:
    df = df[df["method"] != "LightRAG"].reset_index(drop=True)

colors = {
    "Vanilla RAG": "#EDD896",
    "MS-GraphRAG": "#9FD4B0",
    "HippoRAG": "#62CFC8",
    "GFM-RAG": "#E5A070",
    "LightRAG": "#C9B5E8",
}

plt.rcParams.update(
    {
        "font.family": "serif",
        "font.serif": ["Times New Roman", "Times", "DejaVu Serif"],
        "mathtext.fontset": "stix",
    }
)

fig, ax = plt.subplots(figsize=(2.78, 2.28), dpi=300)
fig.patch.set_facecolor("white")
ax.set_facecolor("white")

size_scale = 118

ax.set_xlim(67, 86.5)
ax.set_ylim(20, 80)
ax.set_xticks([70, 75, 80, 85])
ax.set_yticks([20, 30, 40, 50, 60, 70, 80])
ax.grid(True, color="#D0D0D0", linewidth=0.5, zorder=0)
ax.set_axisbelow(True)

for _, row in df.iterrows():
    ax.scatter(
        row["recall"],
        row["relevance"],
        s=size_scale * (row["llm_acc"] - 50) ** 2,
        color=colors[row["method"]],
        alpha=0.36,
        edgecolor="#4A4A4A",
        linewidth=0.5,
        zorder=2,
    )

for _, row in df.iterrows():
    x, y = row["recall"], row["relevance"]

    ax.scatter(
        x,
        y,
        marker="+",
        color="black",
        s=16,
        linewidth=0.75,
        zorder=5,
    )

    ax.annotate(
        f"{row['method']}\n({row['recall']:.1f} / {row['relevance']:.1f})",
        xy=(x, y),
        xytext=(0, 5),
        textcoords="offset points",
        ha="center",
        va="bottom",
        fontsize=5.0,
        zorder=6,
    )

ax.text(
    80.3,
    73.5,
    r"Size represents LLM-ACC" + "\n" + r"$(Diameter\ \propto\ ACC - 50)$",
    ha="center",
    va="center",
    fontsize=9.2,
    fontstyle="italic",
    linespacing=1.02,
    zorder=7,
    bbox={
        "boxstyle": "round,pad=0.38,rounding_size=0.30",
        "facecolor": "white",
        "edgecolor": "black",
        "linewidth": 0.5,
    },
)

ax.set_xlabel("Recall (%)", fontsize=9, fontweight="bold")
ax.set_ylabel("Relevance (%)", fontsize=9, fontweight="bold")
ax.tick_params(labelsize=8)

for spine in ax.spines.values():
    spine.set_linewidth(0.8)
    spine.set_color("black")

plt.tight_layout(pad=0.2)
plt.savefig("rag_bubble_plot.png", dpi=300, bbox_inches="tight", facecolor="white")
plt.savefig("rag_bubble_plot.pdf", bbox_inches="tight", facecolor="white")
plt.show()