Resolve DOI / Title Precisely
MetadataRetrievalBeginner

Resolve DOI / Title Precisely

Quickly fetch metadata, full text, and citation evidence as a basic entry point for research agents

View on GitHub
Scenario
Users already have a DOI or paper title and only need to fetch metadata, full text, and citation evidence quickly. This is a basic operation for research agents.
Estimated calls
~3-5 API calls per resolution
Tools
meta-searchcontent
Pipeline
DOI/标题→ meta-search 精确查询→ 元数据→ content 全文

Input example

Input:
DOI: 10.xxxx/example
or title: "Attention is All You Need"

Output example

Resolved paper:
- title: ...
- doc_id: ...
- year: ...
- venue: ...
- available content offset: 0

Agent Prompt example

You are a DOI and title resolver. Use structured search to locate the exact paper, then read metadata and a short content preview for verification.

Implementation steps

Step 1: Set up the environment

Configure token and HTTP client

!pip install httpx
import os
os.environ["SCIVERSE_API_TOKEN"] = "sv-your-token-here"  # 替换为你的真实值

Step 2: Find a paper precisely by DOI

Use meta-search filters or precise title matching to locate the paper

import os
import asyncio
import httpx

BASE = "https://api.sciverse.space"
TOKEN = os.environ["SCIVERSE_API_TOKEN"]
HEADERS = {"Authorization": f"Bearer {TOKEN}"}

async def resolve_doi(doi: str):
    """\u901a\u8fc7 DOI \u7cbe\u786e\u67e5\u627e\u6587\u732e\u5143\u6570\u636e"""
    async with httpx.AsyncClient(timeout=30) as client:
        resp = await client.post(
            f"{BASE}/meta-search", headers=HEADERS,
            json={
                "query": doi,
                "filters": [{"field": "doi", "operator": "FILTER_OP_EQ", "value": doi}],
                "page": 1, "page_size": 1
            }
        )
        resp.raise_for_status()
        data = resp.json()
        if data.get("total_count", 0) > 0:
            return (data.get("results") or [None])[0]
        return None

async def resolve_title(title: str):
    """\u901a\u8fc7\u6807\u9898\u6a21\u7cca\u67e5\u627e"""
    async with httpx.AsyncClient(timeout=30) as client:
        resp = await client.post(
            f"{BASE}/meta-search", headers=HEADERS,
            json={"query": title, "filters": [], "page": 1, "page_size": 5}
        )
        resp.raise_for_status()
        return (resp.json().get("results") or [])

# \u793a\u4f8b\uff1a\u901a\u8fc7 DOI \u89e3\u6790
paper = await resolve_doi("10.1038/s41586-021-03819-2")
if paper:
    print(f"Title: {paper['title']}")
    print(f"Venue: {paper.get('publication_venue_name_unified', 'N/A')}")
    print(f"Year: {paper.get('publication_published_year', 'N/A')}")
    print(f"Doc ID: {paper.get('doc_id', 'N/A')}")

Step 3: Read a full-text preview

Call content to read the abstract or opening section

async def read_abstract(doc_id: str):
    """\u8bfb\u53d6\u8bba\u6587\u5f00\u5934 1500 \u5b57\u7b26\u4f5c\u4e3a\u6458\u8981"""
    async with httpx.AsyncClient(timeout=30) as client:
        resp = await client.get(
            f"{BASE}/content", headers=HEADERS,
            params={"doc_id": doc_id, "offset": 0, "limit": 1500}
        )
        resp.raise_for_status()
        return resp.json()["text"]

if paper and paper.get("doc_id"):
    abstract = await read_abstract(paper["doc_id"])
    print(f"\
Full text preview:\
{abstract[:500]}...")

Notes

  • Prefer DOI when available because it is more precise than title matching.
  • Normalize punctuation and casing in titles.
  • Return multiple candidates when exact matching is ambiguous.
  • Always include doc_id for downstream calls.

FAQ

DOI 查不到怎么办?

可以回退到标题、作者和年份组合检索。

解析结果要返回什么?

至少返回 title、authors、year、venue、doi、doc_id。

PMID 和 DOI 都有时用哪个?

可同时保留，DOI 适合跨出版商识别，PMID 适合生物医学场景。

解析后下一步做什么?

通常用 doc_id 调 content 读取摘要或全文上下文。

Next steps

Need an API key?

Create one in Console > Tokens.The same API key works for enabled Sciverse, DianShi, and Skills capabilities, with starter quota available according to account permissions.

Open console