data-driven listing and abstract pages for research papers from arxiv…

…, also includes github repos where available

package.json

@@ -7,16 +7,22 @@
     "start": "astro dev",
     "build": "astro check && astro build",
     "preview": "astro preview",
-    "astro": "astro"
+    "astro": "astro",
+    "test": "vitest"
   },
   "dependencies": {
     "@astrojs/check": "^0.9.3",
     "@astrojs/starlight": "^0.26.1",
     "@astrojs/starlight-tailwind": "^2.0.3",
     "@astrojs/tailwind": "^5.1.2",
+    "@effect/platform-node": "^0.65.0",
     "astro": "^4.14.5",
+    "date-fns": "^4.1.0",
+    "effect": "^3.11.0",
     "sharp": "^0.33.5",
     "tailwindcss": "^3.4.15",
-    "typescript": "^5.5.4"
+    "typescript": "^5.5.4",
+    "vitest": "^2.1.7",
+    "xml-js": "^1.6.11"
   }
 }

src/components/GrArxivPage.astro

@@ -0,0 +1,82 @@
+
+---
+import { format } from 'date-fns';
+
+import {extractArxivID, getPaper} from '../lib/arxiv'
+
+import StarlightPage from '@astrojs/starlight/components/StarlightPage.astro';
+
+const props = Astro.props;
+
+const arxivEntry = props.arxiv
+
+const arxivid = extractArxivID(arxivEntry.id)
+
+const paper = getPaper(arxivid)
+
+---
+<StarlightPage 
+  frontmatter={{ title: arxivEntry.title, editUrl:false, tableOfContents:false} }}
+>
+  <p class="byline text-xs">
+    <time itemprop="published" datetime={format(arxivEntry.published, 'yyyy-MM-dd')}>
+      Published {format(arxivEntry.published, 'MMMM do, yyyy')}
+    </time>
+    <address class="author text-xs">By 
+      {arxivEntry.author.map( (author:any, i:number) => (
+        <span>{(i ? ', ' : '')}<a rel="author" class="url fn n">{author.name}</a></span>
+      ))}
+    </address>
+    <cite class="arxivid text-xs">
+      <a href={arxivEntry.id}>arXiv:{arxivid}</a> 
+      [ {arxivEntry.category.join(", ")}
+      ]
+    </cite>
+    <cite class="github text-xs">
+      {paper?.github !== undefined && paper.github !== "" ?
+        (<a href={"https://github.com/" + paper?.github}>github:{paper?.github}</a>)
+        : ''
+      }
+
+    </cite>
+  </p>
+
+  <h2 id="quote">Abstract</h2>
+  <blockquote><p>{arxivEntry.summary}</p></blockquote>
+
+</StarlightPage>
+
+
+
+<!-- <StarlightPage 
+  frontmatter={{ title: `${entry.title.text}`, editUrl:false, tableOfContents:false} }}
+>
+
+<p class="byline text-xs">
+  <time pubdate={format(pubDate, 'yyyy-MM-dd')} title="August 28th, 2011">Published {format(pubDate, 'MMMM do, yyyy')}</time>
+  <address class="author text-xs">By 
+    {authors.map( (author:any, i:number) => (
+      <span key={i}>{(i ? ', ' : '')}
+        <a rel="author" class="url fn n">{author.name.text}</a>
+      </span>
+    ))}
+  </address>
+  <cite class="arxivid text-xs">
+    <a href={entry.id.text}>arXiv:{arxivid}</a> 
+    [ {categories.map( (category:any) => category._attributes.term).join(", ")}
+    ]
+  </cite>
+  <cite class="github text-xs">
+    {paper?.github !== undefined && paper.github !== "" ?
+      (<a href={"https://github.com/" + paper?.github}>github:{paper?.github}</a>)
+      : ''
+    }
+    
+  </cite>
+</p>
+
+
+<h2 id="quote">Abstract</h2>
+<blockquote><p>{entry.summary.text}</p></blockquote>
+
+</StarlightPage> -->

src/content/config.ts

@@ -1,6 +1,15 @@
-import { defineCollection } from 'astro:content';
+import { z, defineCollection } from 'astro:content';
 import { docsSchema } from '@astrojs/starlight/schema';
 
+const researchCollection = defineCollection({
+	type: 'data',
+	schema: z.object({
+		title: z.string(),
+		canonicalURL: z.string().url()
+	})
+})
+
 export const collections = {
 	docs: defineCollection({ schema: docsSchema() }),
+	research: researchCollection
 };

src/content/docs/appendices/notation.md

@@ -12,11 +12,11 @@ The GraphRAG pattern catalog uses a data notation called `gram` to describe
 logical graph structures called patterns that are composed of nodes, relationships 
 and subjects.
 
-The Gram notation is intended to be self-descriptive and explicit, able to 
-represent data and structures that are often implicit in a physical graph models. 
-For example, paths are present in any connected graph, however storing path-level 
-information isn't normally supported. You can find paths, even store paths, but 
-there is no way to "say something" about a path.
+> The Gram notation is intended to be self-descriptive and explicit, able to 
+> represent data and structures that are often implicit in a physical graph models. 
+> For example, paths are present in any connected graph, however storing path-level 
+> information isn't normally supported. You can find paths, even store paths, but 
+> there is no way to "say something" about a path.
 
 Gram starts with a notion of "subjects" as a self-describing data structure
 in two parts:

src/content/docs/appendices/research.mdx

@@ -1,6 +1,8 @@
 ---
 title: GraphRAG Papers
 description: Foundational research papers about GraphRAG and Knowledge Graphs
+editUrl: false
+tableOfContents: false
 ---
 
 - [Graph Retrieval-Augmented Generation: A Survey](https://arxiv.org/abs/2408.08921)

src/data/papers.json

@@ -0,0 +1,86 @@
+[
+  {
+    "arxivid": "2408.08921",
+    "github": "pengboci/GraphRAG-Survey"
+  },
+  {
+    "arxivid": "2312.16890",
+    "github": "HKUDS/DiffKG"
+  },
+  {
+    "arxivid": "2306.08302",
+    "github": ""
+  },
+  {
+    "arxivid": "2310.04560",
+    "github": "google-research/talk-like-a-graph"
+  },
+  {
+    "arxivid": "2311.07509",
+    "github": "datadotworld/cwd-benchmark-data"
+  },
+  {
+    "arxivid": "2402.07630",
+    "github": "XiaoxinHe/G-Retriever"
+  },
+  {
+    "arxivid": "2404.12491",
+    "github": "urchade/GraphER"
+  },
+  {
+    "arxivid": "2404.16130",
+    "github": "microsoft/graphrag"
+  },
+  {
+    "arxivid": "2404.17723",
+    "github": ""
+  },
+  {
+    "arxivid": "2408.04948",
+    "github": ""
+  },
+  {
+    "arxivid": "2406.14550",
+    "github": ""
+  },
+  {
+    "arxivid": "2410.05779",
+    "github": "HKUDS/LightRAG"
+  },
+  {
+    "arxivid": "2410.08815",
+    "github": "Li-Z-Q/StructRAG"
+  },
+  {
+    "arxivid": "2307.07697",
+    "github": "IDEA-FinAI/ToG"
+  },
+  {
+    "arxivid": "2405.14831",
+    "github": "OSU-NLP-Group/HippoRAG"
+  },
+  {
+    "arxivid": "2408.04187",
+    "github": "MedicineToken/Medical-Graph-RAG"
+  },
+  {
+    "arxivid": "2405.18414",
+    "github": ""
+  },
+  {
+    "arxivid": "2405.16506",
+    "github": "HuieL/GRAG"
+  },
+  {
+    "arxivid": "2410.23875",
+    "github": "liyichen-cly/PoG"
+  },
+  {
+    "arxivid": "2410.18415",
+    "github": ""
+  },
+  {
+    "arxivid": "2404.07103",
+    "github": "PeterGriffinJin/Graph-CoT"
+  }
+]

src/lib/arxiv.spec.ts

@@ -0,0 +1,48 @@
+import { assert, expect, test } from 'vitest';
+
+import { Effect, Either } from "effect"
+import { FetchHttpClient } from "@effect/platform"
+
+import { getArxivDetails } from './arxiv'
+import { type ArxivEntry } from './arxiv'
+
+test('arxiv fetch well-known entry', async () => {
+	const arxivid = '2402.07630';
+
+	const program = getArxivDetails(arxivid).pipe(
+    Effect.scoped,
+    Effect.provide(FetchHttpClient.layer) // provide a real implementation of fetch() 
+  );
+
+  const result = await Effect.runPromise(program)
+
+  expect(Either.isRight(result))
+
+  expect(Either.getOrThrow(result).title).toBe("G-Retriever: Retrieval-Augmented Generation for Textual Graph Understanding and Question Answering")
+
+  // DEBUG
+  // Either.match(result,
+  //   {
+  //     onLeft: (e) => console.error(e),
+  //     onRight: (result) => console.log(result.feed.entry[0])
+  //   }
+  // )
+
+});
+
+test('arxiv get many entries', async () => {
+	const arxivids = ['2402.07630', '2311.07509', '2306.08302'];
+
+  const program = Effect.forEach(arxivids, (arxivid) =>
+    getArxivDetails(arxivid)
+  ).pipe(
+    Effect.map( Either.all ), // gather all the successes into one array
+    Effect.scoped,
+    Effect.provide(FetchHttpClient.layer)
+  )
+
+  const result = await Effect.runPromise(program)
+
+  expect(Either.isRight(result))
+
+})