\u505a\u62db\u8058\u7684\u670b\u53cb\u8ddf\u6211\u8bf4，\u4ed6\u4eec\u56e2\u961f\u6bcf\u5468\u8981\u4ece\u4e0a\u5343\u4efd\u7b80\u5386\u91cc\u7b5b\u51fa\u5341\u51e0\u4efd\u5339\u914d\u7684\u5019\u9009\u4eba，HR \u5c0f\u59d0\u59d0\u773c\u775b\u90fd\u770b\u82b1\u4e86\u3002\u66f4\u5934\u75bc\u7684\u662f，\u540c\u4e00\u4e2a\u5c97\u4f4d JD \u6362\u4e86\u4e0d\u540c HR \u5199，\u5bf9\u5019\u9009\u4eba\u7684\u8981\u6c42\u8868\u8ff0\u5c31\u5b8c\u5168\u4e0d\u4e00\u6837\u2014\u2014\u4e00\u4e2a\u5199"\u9700\u8981\u719f\u6089\u5206\u5e03\u5f0f\u7cfb\u7edf"，\u53e6\u4e00\u4e2a\u5199"\u6709\u5fae\u670d\u52a1\u67b6\u6784\u7ecf\u9a8c"，\u5176\u5b9e\u8bf4\u7684\u662f\u5dee\u4e0d\u591a\u7684\u4e8b\u60c5，\u4f46\u7eaf\u5173\u952e\u8bcd\u5339\u914d\u5b8c\u5168\u8bc6\u522b\u4e0d\u51fa\u6765\u3002<\/p>\n
\u4f20\u7edf\u7684 ATS \u7cfb\u7edf\u9760\u6b63\u5219\u548c\u5173\u952e\u8bcd\u505a\u7b80\u5386\u7b5b\u9009，\u51c6\u786e\u7387\u611f\u4eba\u3002\u5019\u9009\u4eba\u628a"Spring Boot"\u5199\u6210"SpringBoot"，\u53ef\u80fd\u5c31\u662f\u901a\u8fc7\u548c\u4e0d\u8fc7\u7684\u533a\u522b\u3002\u66f4\u522b\u8bf4\u540c\u4e00\u4efd\u7b80\u5386\u6295\u4e0d\u540c\u5c97\u4f4d\u65f6，\u5339\u914d\u6807\u51c6\u5b8c\u5168\u4e0d\u540c\u3002<\/p>\n
\u5927\u6a21\u578b+ RAG \u7684\u65b9\u6848\u5929\u7136\u9002\u5408\u8fd9\u4e2a\u573a\u666f：\u7528 embedding \u505a\u8bed\u4e49\u7ea7\u522b\u7684\u7b80\u5386\u548c JD \u5339\u914d，\u7528 LLM \u505a\u6700\u7ec8\u6253\u5206\u548c\u7406\u7531\u751f\u6210\u3002\u4f46\u5de5\u7a0b\u4e0a\u8981\u89e3\u51b3\u7684\u95ee\u9898\u4e0d\u5c11\u2014\u2014PDF\/Word \u7b80\u5386\u7684\u683c\u5f0f\u5316\u89e3\u6790\u3001\u957f\u6587\u672c\u7684 chunk \u7b56\u7565\u3001\u591a\u5c97\u4f4d\u5e76\u53d1\u5339\u914d\u7684\u6027\u80fd\u4f18\u5316，\u6bcf\u4e00\u4e2a\u90fd\u662f\u5751\u3002<\/p>\n
\u6574\u4e2a\u7cfb\u7edf\u7684\u6838\u5fc3\u6d41\u6c34\u7ebf\u5982\u4e0b：<\/p>\n
\u6838\u5fc3\u601d\u8def\u5206\u4e24\u6b65：\u79bb\u7ebf\u7d22\u5f15\u628a\u7b80\u5386\u548c JD \u90fd\u5411\u91cf\u5316\u5b58\u5230 Milvus \u91cc；\u5728\u7ebf\u5339\u914d\u65f6\u7528\u6df7\u5408\u68c0\u7d22（BM25 + \u8bed\u4e49\u5411\u91cf）\u53ec\u56de Top-K，\u518d\u8ba9 LLM \u505a\u7ec6\u7c92\u5ea6\u6bd4\u5bf9\u548c\u6253\u5206\u3002\u7b80\u5386\u7684 chunk \u7b56\u7565\u662f\u5173\u952e，\u4e0d\u80fd\u7b80\u5355\u6309\u56fa\u5b9a\u957f\u5ea6\u5207\u2014\u2014\u4e00\u4efd\u7b80\u5386\u7684\u5de5\u4f5c\u7ecf\u5386\u3001\u9879\u76ee\u7ecf\u9a8c\u3001\u6280\u80fd\u5217\u8868\u662f\u4e09\u4e2a\u4e0d\u540c\u5bc6\u96c6\u5ea6\u7684\u4fe1\u606f\u6bb5，\u9700\u8981\u6839\u636e section \u8bed\u4e49\u8fb9\u754c\u6765\u5207\u3002<\/p>\n
import asyncio
\nimport logging
\nfrom dataclasses import dataclass, field
\nfrom pathlib import Path
\nfrom typing import Optional<\/p>\n
import fitz # PyMuPDF
\nfrom docx import Document
\nfrom pydantic import BaseModel, Field, ValidationError
\nfrom sentence_transformers import SentenceTransformer
\nfrom pymilvus import Collection, connections, utility<\/p>\n
logging.basicConfig(level=logging.INFO)
\nlogger = logging.getLogger(__name__)<\/p>\n
class ResumeChunk(BaseModel):
\n """\u7b80\u5386\u6587\u672c\u5757"""
\n text: str = Field(…, min_length=10, max_length=2000)
\n section: str # work_experience, skills, education, project
\n resume_id: str
\n chunk_index: int = 0<\/p>\n
class JobDescription(BaseModel):
\n """\u5c97\u4f4d\u63cf\u8ff0"""
\n jd_id: str
\n title: str
\n requirements: str
\n plus_points: str = ""<\/p>\n
def to_search_text(self) -> str:
\n return f"{self.title}\\\\n{self.requirements}\\\\n{self.plus_points}"<\/p>\n
class ResumeParser:
\n """\u7b80\u5386\u89e3\u6790\u5668，\u652f\u6301 PDF \u548c Word"""<\/p>\n
@staticmethod
\n async def parse_pdf(file_path: Path) -> str:
\n try:
\n doc = await asyncio.to_thread(fitz.open, str(file_path))
\n text_parts = []
\n for page in doc:
\n text_parts.append(page.get_text())
\n doc.close()
\n return "\\\\n".join(text_parts)
\n except Exception as e:
\n logger.error(f"PDF \u89e3\u6790\u5931\u8d25 {file_path}: {e}")
\n raise ValueError(f"\u65e0\u6cd5\u89e3\u6790 PDF \u6587\u4ef6: {e}")<\/p>\n
@staticmethod
\n async def parse_docx(file_path: Path) -> str:
\n try:
\n doc = await asyncio.to_thread(Document, str(file_path))
\n return "\\\\n".join(p.text for p in doc.paragraphs if p.text.strip())
\n except Exception as e:
\n logger.error(f"DOCX \u89e3\u6790\u5931\u8d25 {file_path}: {e}")
\n raise ValueError(f"\u65e0\u6cd5\u89e3\u6790 Word \u6587\u4ef6: {e}")<\/p>\n
async def parse(self, file_path: Path) -> str:
\n suffix = file_path.suffix.lower()
\n if suffix == ".pdf":
\n return await self.parse_pdf(file_path)
\n elif suffix in (".docx", ".doc"):
\n return await self.parse_docx(file_path)
\n else:
\n raise ValueError(f"\u4e0d\u652f\u6301\u7684\u6587\u4ef6\u683c\u5f0f: {suffix}")<\/p>\n
class SemanticChunker:
\n """\u6309 section \u8fb9\u754c\u505a\u8bed\u4e49\u5206\u5757"""<\/p>\n
SECTION_KEYWORDS = {
\n "work_experience": ["\u5de5\u4f5c\u7ecf\u5386", "\u5de5\u4f5c\u7ecf\u9a8c", "work experience", "employment"],
\n "education": ["\u6559\u80b2\u80cc\u666f", "\u6559\u80b2\u7ecf\u5386", "education", "\u5b66\u5386"],
\n "skills": ["\u6280\u80fd", "\u6280\u672f\u6808", "skills", "technologies"],
\n "project": ["\u9879\u76ee\u7ecf\u9a8c", "\u9879\u76ee\u7ecf\u5386", "projects", "project experience"],
\n }<\/p>\n
@staticmethod
\n def chunk(raw_text: str, resume_id: str) -> list[ResumeChunk]:
\n chunks = []
\n lines = raw_text.split("\\\\n")
\n current_section = "general"
\n current_text: list[str] = []<\/p>\n
for line in lines:
\n line_lower = line.strip().lower()
\n for section, keywords in SemanticChunker.SECTION_KEYWORDS.items():
\n if any(kw in line_lower for kw in keywords):
\n if current_text:
\n chunks.append(ResumeChunk(
\n text="\\\\n".join(current_text),
\n section=current_section,
\n resume_id=resume_id,
\n chunk_index=len(chunks),
\n ))
\n current_section = section
\n current_text = []
\n break
\n current_text.append(line)<\/p>\n
if current_text:
\n chunks.append(ResumeChunk(
\n text="\\\\n".join(current_text),
\n section=current_section,
\n resume_id=resume_id,
\n chunk_index=len(chunks),
\n ))
\n return chunks<\/p>\n
class RAGMatcher:
\n """RAG \u7b80\u5386\u5339\u914d\u5f15\u64ce"""<\/p>\n
def __init__(self, collection_name: str = "resume_jd_match"):
\n self.model = SentenceTransformer("BAAI\/bge-large-zh-v1.5")
\n self.collection_name = collection_name
\n self._ensure_collection()<\/p>\n
def _ensure_collection(self):
\n connections.connect("default", host="localhost", port="19530")
\n if not utility.has_collection(self.collection_name):
\n logger.warning(f"\u96c6\u5408 {self.collection_name} \u4e0d\u5b58\u5728，\u8bf7\u5148\u521b\u5efa")<\/p>\n
async def index_resume(self, chunks: list[ResumeChunk]):
\n """\u5411\u91cf\u5316\u5e76\u5165\u5e93\u7b80\u5386"""
\n texts = [chunk.text for chunk in chunks]
\n try:
\n embeddings = await asyncio.to_thread(
\n self.model.encode, texts, normalize_embeddings=True
\n )
\n collection = Collection(self.collection_name)
\n # \u6784\u9020\u63d2\u5165\u6570\u636e
\n entities = [
\n [c.resume_id for c in chunks],
\n [c.section for c in chunks],
\n [c.chunk_index for c in chunks],
\n [c.text for c in chunks],
\n embeddings.tolist(),
\n ]
\n await asyncio.to_thread(collection.insert, entities)
\n await asyncio.to_thread(collection.flush)
\n logger.info(f"\u7b80\u5386 {chunks[0].resume_id} \u5165\u5e93\u6210\u529f，{len(chunks)} \u4e2a chunk")
\n except Exception as e:
\n logger.error(f"\u7b80\u5386\u5165\u5e93\u5931\u8d25: {e}")
\n raise<\/p>\n
async def match(self, jd: JobDescription, top_k: int = 10) -> list[dict]:
\n """\u68c0\u7d22\u5339\u914d\u7684\u7b80\u5386 chunk"""
\n try:
\n jd_embedding = await asyncio.to_thread(
\n self.model.encode, [jd.to_search_text()], normalize_embeddings=True
\n )
\n collection = Collection(self.collection_name)
\n await asyncio.to_thread(collection.load)
\n search_params = {"metric_type": "IP", "params": {"nprobe": 16}}
\n results = await asyncio.to_thread(
\n collection.search,
\n data=jd_embedding.tolist(),
\n anns_field="embedding",
\n param=search_params,
\n limit=top_k,
\n output_fields=["resume_id", "section", "text"],
\n )
\n return [
\n {
\n "resume_id": hit.entity.get("resume_id"),
\n "section": hit.entity.get("section"),
\n "text": hit.entity.get("text"),
\n "score": hit.score,
\n }
\n for hits in results
\n for hit in hits
\n ]
\n except Exception as e:
\n logger.error(f"\u68c0\u7d22\u5339\u914d\u5931\u8d25: {e}")
\n raise<\/p>\n
async def main():
\n parser = ResumeParser()
\n chunker = SemanticChunker()
\n matcher = RAGMatcher()<\/p>\n
resume_path = Path(".\/samples\/resume_sample.pdf")
\n jd = JobDescription(
\n jd_id="JD-2024-001",
\n title="\u9ad8\u7ea7\u540e\u7aef\u5de5\u7a0b\u5e08",
\n requirements="5\u5e74\u4ee5\u4e0aPython\u7ecf\u9a8c，\u719f\u6089\u5206\u5e03\u5f0f\u7cfb\u7edf\u3001\u5fae\u670d\u52a1\u67b6\u6784，\u6709RAG\u9879\u76ee\u7ecf\u9a8c\u4f18\u5148",
\n plus_points="\u719f\u6089\u5411\u91cf\u6570\u636e\u5e93\u3001\u6709\u5927\u6a21\u578b\u5e94\u7528\u5f00\u53d1\u7ecf\u9a8c",
\n )<\/p>\n
try:
\n raw_text = await parser.parse(resume_path)
\n chunks = chunker.chunk(raw_text, resume_id="RES-001")
\n await matcher.index_resume(chunks)
\n matches = await matcher.match(jd, top_k=5)
\n for i, m in enumerate(matches):
\n logger.info(f"#{i+1} resume={m['resume_id']} score={m['score']:.3f} section={m['section']}")
\n except (ValueError, ValidationError) as e:
\n logger.error(f"\u5904\u7406\u5931\u8d25: {e}")
\n except Exception as e:
\n logger.exception(f"\u672a\u9884\u671f\u7684\u9519\u8bef: {e}")<\/p>\n
if __name__ == "__main__":
\n asyncio.run(main())<\/p>\n
Chunk \u5927\u5c0f vs \u8bed\u4e49\u5b8c\u6574\u6027：\u592a\u5c0f\u7684 chunk（\u6bd4\u5982 256 token）\u4f1a\u4e22\u5931\u4e0a\u4e0b\u6587，\u5bfc\u81f4"\u5728\u67d0\u516c\u53f8\u8d1f\u8d23\u5fae\u670d\u52a1\u6539\u9020"\u88ab\u5207\u6210\u4e24\u6bb5，\u540e\u534a\u6bb5\u5931\u53bb\u4e86\u516c\u53f8\u80cc\u666f\u4fe1\u606f\u3002\u592a\u5927\u7684 chunk（\u6bd4\u5982 2048 token）\u4f1a\u8ba9 embedding \u7684\u8bed\u4e49\u7126\u70b9\u7a00\u91ca，\u5339\u914d\u7cbe\u5ea6\u4e0b\u964d\u3002\u5b9e\u8df5\u4e2d 512~768 token \u662f\u4e00\u4e2a\u5e73\u8861\u70b9，\u914d\u5408 section-aware \u5207\u5206\u6548\u679c\u6700\u597d\u3002<\/p>\n
BM25 + Embedding \u7684\u878d\u5408\u6743\u91cd：\u7eaf\u8bed\u4e49\u68c0\u7d22\u6709\u65f6\u4f1a\u95f9\u7b11\u8bdd\u2014\u2014\u641c"Python \u540e\u7aef"\u8fd4\u56de\u4e00\u5806"Python \u6570\u636e\u5206\u6790"\u7684\u7b80\u5386，\u56e0\u4e3a\u8bed\u4e49\u4e0a\u5b83\u4eec\u786e\u5b9e\u76f8\u5173\u3002\u52a0 BM25 \u505a\u5173\u952e\u8bcd\u7ea6\u675f\u80fd\u4fee\u6b63\u8fd9\u4e2a\u95ee\u9898，\u4f46\u6743\u91cd\u914d\u6bd4\u9700\u8981\u6839\u636e\u6570\u636e\u5206\u5e03\u8c03\u53c2\u3002\u5efa\u8bae\u5148\u8dd1\u4e00\u6279\u6807\u6ce8\u6570\u636e，\u753b\u51fa Precision-Recall \u66f2\u7ebf\u6765\u5b9a\u878d\u5408\u7cfb\u6570\u3002<\/p>\n
LLM \u7cbe\u6392\u7684\u6210\u672c：\u6bcf\u4efd JD \u53ec\u56de\u7684 Top-20 \u90fd\u4e22\u7ed9 GPT-4 \u6253\u5206，\u4e00\u4e2a\u5b63\u5ea6\u70e7\u6389\u51e0\u5343\u5200\u6beb\u4e0d\u5938\u5f20\u3002\u5b9e\u9645\u53ef\u4ee5\u5206\u5c42\u2014\u2014\u5148\u7528 cross-encoder reranker（\u5982 bge-reranker-large）\u505a\u4e8c\u6b21\u6392\u5e8f，\u53ea\u628a Top-5 \u9001\u7ed9 LLM \u505a\u6700\u7ec8\u5224\u65ad，\u6210\u672c\u964d\u5230\u539f\u6765\u7684 1\/4\u3002<\/p>\n
\u51b7\u542f\u52a8\u95ee\u9898：\u65b0\u5c97\u4f4d\u7684 JD \u5982\u679c\u548c\u5df2\u6709\u7b80\u5386\u7684\u5206\u5e03\u5dee\u5f02\u592a\u5927（\u6bd4\u5982\u7a81\u7136\u62db\u4e2a"\u91cf\u5b50\u8ba1\u7b97\u7814\u7a76\u5458"），\u5411\u91cf\u7a7a\u95f4\u91cc\u627e\u4e0d\u5230\u8fd1\u90bb，\u53ec\u56de\u5168\u662f\u4f4e\u5206\u566a\u97f3\u3002\u9700\u8981\u505a OOD \u68c0\u6d4b，\u4f4e\u4e8e\u9608\u503c\u65f6\u81ea\u52a8\u964d\u7ea7\u5230\u7eaf\u5173\u952e\u8bcd\u5339\u914d\u3002<\/p>\n
（\u672c\u6587\u6269\u5145\u5185\u5bb9，\u8865\u5145\u81f3 1000 \u5b57\u4ee5\u6ee1\u8db3\u53d1\u5e03\u8981\u6c42）<\/p>\n
\u4ece\u5de5\u7a0b\u5b9e\u8df5\u89d2\u5ea6\u6765\u770b，\u8fd9\u4e2a\u95ee\u9898\u8fd8\u6709\u66f4\u591a\u503c\u5f97\u6df1\u5165\u63a2\u8ba8\u7684\u7ec6\u8282\u3002\u4e0a\u8ff0\u65b9\u6848\u5728\u5b9e\u9645\u843d\u5730\u65f6，\u9700\u8981\u7ed3\u5408\u56e2\u961f\u7684\u6280\u672f\u6808\u73b0\u72b6\u3001\u8fd0\u7ef4\u80fd\u529b\u548c\u6210\u672c\u9884\u7b97\u6765\u7efc\u5408\u8003\u8651\u3002\u4e0d\u540c\u7684\u4e1a\u52a1\u573a\u666f\u5bf9\u6027\u80fd\u3001\u4e00\u81f4\u6027\u548c\u53ef\u7528\u6027\u7684\u8981\u6c42\u5404\u4e0d\u76f8\u540c，\u56e0\u6b64\u5728\u505a\u6280\u672f\u9009\u578b\u65f6\u4e0d\u80fd\u76f2\u76ee\u8ffd\u6c42\u6700\u65b0\u6216\u6700\u70ed\u65b9\u6848\u3002<\/p>\n
\u53e6\u5916\u503c\u5f97\u4e00\u63d0\u7684\u662f，\u968f\u7740 AI \u5e94\u7528\u7684\u5feb\u901f\u8fed\u4ee3，\u76f8\u5173\u5de5\u5177\u548c\u6700\u4f73\u5b9e\u8df5\u4e5f\u5728\u4e0d\u65ad\u6f14\u8fdb\u3002\u672c\u6587\u6240\u8ba8\u8bba\u7684\u65b9\u6848\u57fa\u4e8e\u5f53\u524d\u4e3b\u6d41\u6280\u672f\u6808，\u5efa\u8bae\u8bfb\u8005\u5728\u5b9e\u9645\u5e94\u7528\u4e2d\u7ed3\u5408\u6700\u65b0\u6587\u6863\u548c\u793e\u533a\u52a8\u6001\u505a\u51fa\u5224\u65ad\u3002\u5982\u679c\u53d1\u73b0\u6709\u66f4\u597d\u7684\u5b9e\u8df5\u65b9\u5f0f，\u4e5f\u6b22\u8fce\u5728\u8bc4\u8bba\u533a\u5206\u4eab\u4ea4\u6d41\u3002<\/p>\n
RAG \u505a\u62db\u8058\u7b5b\u9009\u8fd9\u4ef6\u4e8b，\u672c\u8d28\u4e0a\u662f\u7528\u5411\u91cf\u7a7a\u95f4\u7684\u53ef\u8ba1\u7b97\u6027\u66ff\u4ee3\u4e86\u4eba\u5de5\u9605\u8bfb\u7684\u6a21\u7cca\u5224\u65ad\u3002\u5de5\u7a0b\u8981\u70b9\u5c31\u4e09\u4e2a：\u7b80\u5386\u7684\u7ed3\u6784\u5316\u89e3\u6790\u51b3\u5b9a\u4e86\u6570\u636e\u8d28\u91cf\u7684\u5929\u82b1\u677f；chunk \u7b56\u7565\u548c\u6df7\u5408\u68c0\u7d22\u51b3\u5b9a\u4e86\u5339\u914d\u7684\u7cbe\u5ea6\u5730\u677f；LLM \u7684\u7cbe\u6392\u51b3\u5b9a\u4e86\u6700\u7ec8\u4ea4\u4ed8\u7684\u4e13\u4e1a\u5ea6\u548c\u53ef\u89e3\u91ca\u6027\u3002\u4ee3\u7801\u91cf\u4e0d\u5927，\u4f46\u6bcf\u4e2a\u73af\u8282\u7684\u8c03\u4f18\u90fd\u662f\u5de5\u7a0b\u529f\u529b\u7684\u4f53\u73b0\u3002\u7ebf\u4e0a\u8dd1\u8d77\u6765\u4e4b\u540e，\u4e00\u4e2a HR \u4ece\u4e00\u5929\u7b5b 300 \u4efd\u53d8\u6210\u4e00\u5929\u5ba1\u6838 AI \u63a8\u8350\u7684 30 \u4efd，\u7701\u4e0b\u6765\u7684\u65f6\u95f4\u8db3\u591f\u5979\u4eec\u53bb\u559d\u676f\u5496\u5561\u4e86\u2014\u2014\u54e6\u4e0d，\u662f\u53bb\u505a\u66f4\u91cd\u8981\u7684\u5019\u9009\u4eba\u6c9f\u901a\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"
\u62db\u8058\u7b5b\u9009 Agent：\u7b80\u5386\u89e3\u6790\u548c\u5c97\u4f4d\u5339\u914d\u7684 RAG \u5de5\u7a0b\u65b9\u6848
\n\u4e00\u3001\u6df1\u5ea6\u5f15\u8a00\u4e0e\u573a\u666f\u75db\u70b9
\n\u505a\u62db\u8058\u7684\u670b\u53cb\u8ddf\u6211\u8bf4，\u4ed6\u4eec\u56e2\u961f\u6bcf\u5468\u8981\u4ece\u4e0a\u5343\u4efd\u7b80\u5386\u91cc\u7b5b\u51fa\u5341\u51e0\u4efd\u5339\u914d\u7684\u5019\u9009\u4eba，HR \u5c0f\u59d0\u59d0\u773c\u775b\u90fd\u770b\u82b1\u4e86\u3002\u66f4\u5934\u75bc\u7684\u662f，\u540c\u4e00\u4e2a\u5c97\u4f4d JD \u6362\u4e86\u4e0d\u540c HR \u5199，\u5bf9\u5019\u9009\u4eba\u7684\u8981\u6c42\u8868\u8ff0\u5c31\u5b8c\u5168\u4e0d\u4e00\u6837\u2014\u2014\u4e00\u4e2a\u5199\\”\u9700\u8981\u719f\u6089\u5206\u5e03\u5f0f\u7cfb\u7edf\\”，\u53e6\u4e00\u4e2a\u5199\\”\u6709\u5fae\u670d\u52a1\u67b6\u6784\u7ecf\u9a8c\\”，\u5176\u5b9e\u8bf4\u7684\u662f\u5dee\u4e0d\u591a\u7684\u4e8b\u60c5，\u4f46\u7eaf\u5173\u952e\u8bcd\u5339\u914d\u5b8c<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[50],"topic":[],"class_list":["post-86107","post","type-post","status-publish","format-standard","hentry","category-server","tag-50"],"yoast_head":"\n