MCPcopy Create free account

hub / github.com/Jennyee1/AcademicAgent / functions

Functions1,438 in github.com/Jennyee1/AcademicAgent

↓ 2 callersFunction_run_one_cell
单 cell:(slug, chunk_size, model) -> result dict.
scripts/experiments/chunk_model_sweep.py:60
↓ 2 callersMethod_save
保存注册表到 JSON
src/core/paper_registry.py:554
↓ 2 callersFunction_save_report
保存报告为 JSON + HTML
skills/learning_path/scripts/analyze_knowledge.py:244
↓ 2 callersFunction_sha256_text
(text: str)
src/evaluation/dataset.py:84
↓ 2 callersFunction_short
(v: Any, max_len: int = 80)
src/runtime/workflow.py:306
↓ 2 callersMethod_split_text
将长文本分段 按段落边界分割,每段不超过 max_chunk_chars 字符。 当前默认每段大约 500 个英文词左右。
src/knowledge/extractor.py:632
↓ 2 callersFunction_trace_excerpt
(traces: list[TraceEvent], task_id: str, limit: int = 4)
src/evaluation/failure_cards.py:148
↓ 2 callersFunction_write_jsonl_append
(path: Path, record: dict)
scripts/experiments/chunk_model_sweep.py:134
↓ 2 callersFunctionaccumulate_evidence
把新证据追加到 edge.evidence;返回累积后的证据数。 去重规则:同 source_paper + source_section 已存在 → 不重复追加(取较高 confidence)。
src/knowledge/evidence.py:90
↓ 2 callersMethodaddItems
(e,t)
lib/tom-select/tom-select.complete.min.js:246
↓ 2 callersMethodaddOptions
(e,t=!1)
lib/tom-select/tom-select.complete.min.js:224
↓ 2 callersMethodadvanceSelection
(e,t)
lib/tom-select/tom-select.complete.min.js:288
↓ 2 callersMethodanalyze_temporal
基于双时态字段的综合分析 利用两个独立时间轴: - first_seen_year(有效时间):知识在现实中何时出现 - created_at(事务时间):知识何时被录入图谱 分析三个维度: 1.
src/knowledge/graph_analyzer.py:344
↓ 2 callersFunctionapply_metric
按名字应用一个指标。未知指标名抛 KeyError。
src/evaluation/metrics/__init__.py:312
↓ 2 callersFunctionbE
(t,e,i,n)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionbg
(t)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionbm
(t)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctioncO
(t,e)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersMethodcanCreate
(e)
lib/tom-select/tom-select.complete.min.js:298
↓ 2 callersFunctioncb
(t,e)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersMethodclose
释放 PDF 文档资源
src/core/pdf_parser.py:134
↓ 2 callersFunctioncoerce_relation_type
把字符串解析为 RelationType。返回 (类型, is_fallback)。
src/knowledge/schema.py:213
↓ 2 callersMethoddisable
()
lib/tom-select/tom-select.complete.min.js:292
↓ 2 callersFunctionestimate_cost
根据 token 数与价格表估算成本。
src/evaluation/cost.py:47
↓ 2 callersMethodextract_all_images
提取所有页面的图片 【工程说明】这里仍返回 list 而非 generator,因为: 1. 下游 cmd_deep() 需要 len() 统计总数 2. 嵌入图片的 base64 通常远小于全页渲染(几十 KB vs 几 MB)
src/core/pdf_parser.py:427
↓ 2 callersFunctionfb
(t,e,i,n,o)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersMethodformat_duplicate_warning
生成去重警告信息(供 Workflow 使用)
src/core/paper_registry.py:509
↓ 2 callersMethodfrom_dict
(cls, d: dict)
src/evaluation/dataset.py:67
↓ 2 callersFunctiongap_type_match
检测出的盲区类型集合对期望类型的覆盖率(seed graph 是固定 fixture,故确定性)。
src/evaluation/metrics/capability.py:64
↓ 2 callersFunctiongenerate_html_report
(run_dir: str | Path)
src/evaluation/reporting/html.py:58
↓ 2 callersFunctiongenerate_markdown_report
(run_dir: str | Path)
src/evaluation/reporting/markdown.py:19
↓ 2 callersMethodgetItem
(e)
lib/tom-select/tom-select.complete.min.js:244
↓ 2 callersMethodgetSearchOptions
()
lib/tom-select/tom-select.complete.min.js:204
↓ 2 callersFunctionget_adapter
按工具名取 adapter;首次调用时懒加载所有 adapter 模块。
src/evaluation/adapters/base.py:99
↓ 2 callersFunctionget_paper_details
获取一篇论文的完整详细信息。需要先通过 search_papers 获取 Paper ID。 适合使用的场景: - 用户想深入了解搜索结果中的某篇论文 - 需要获取论文的完整摘要、PDF 链接或引用信息 Args: paper_id: S
src/mcp_servers/paper_search.py:339
↓ 2 callersFunctionhT
(t,e)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionhb
(t)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctioniT
(t,e)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctioniy
(t)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionjx
(t,e)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionkb
(t,e)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionkg_edge_f1
(src_label, relation_type, tgt_label) 三元组的 F1。 extracted_edges : [{"source_id", "target_id", "relation_type", ...}] gold_edges : [{"sour
src/evaluation/metrics/kg.py:79
↓ 2 callersFunctionky
(t,e,i)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionlatency_stats
按 tool_name 统计 {mean, median, p50, p90, max, count},含 __all__ 聚合。
src/evaluation/metrics/common.py:70
↓ 2 callersFunctionlo
(t)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersMethodload_events
读取 JSONL 文件中所有事件。
src/evaluation/tracer.py:123
↓ 2 callersMethodlock
()
lib/tom-select/tom-select.complete.min.js:292
↓ 2 callersFunctionm
()
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionmain
(argv: list[str] | None = None)
src/recorder/cli.py:180
↓ 2 callersFunctionmain
(argv: list[str] | None = None)
src/runtime/cli.py:145
↓ 2 callersFunctionmark_conflict
在 edge.properties 标记冲突 + 对端 edge_id。
src/knowledge/evidence.py:209
↓ 2 callersFunctionmb
(t,e)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionmg
(t,e)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionmk
(t)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionndcg_at_k
nDCG@k,二元相关性(命中=1)。
src/evaluation/metrics/retrieval.py:85
↓ 2 callersMethodnormalize_label
Layer 1: 用 AliasMap 把 mention 规范化为 canonical 形式。 不查询 store;纯字符串变换。
src/knowledge/entity_linker.py:191
↓ 2 callersFunctionoT
(t,e,i,n)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersMethodonFocus
(e)
lib/tom-select/tom-select.complete.min.js:177
↓ 2 callersFunctionpr
(t)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionprecision_at_k
Precision@k = top-k 中命中数 / k。
src/evaluation/metrics/retrieval.py:45
↓ 2 callersMethodpreload
()
lib/tom-select/tom-select.complete.min.js:189
↓ 2 callersFunctionpy
(t,e)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionrecall_at_k
Recall@k = top-k 中命中的不同 gold 数 / gold 总数。
src/evaluation/metrics/retrieval.py:25
↓ 2 callersFunctionrecord_run_to_refactor
把 run 摘要写到 RefactorRecorder(area="eval-runs")。 与"重构步骤"共享同一记录管道: - log_step:记录这次 run 是"哪个数据集/模型/git hash + 总数 + 关键指标" - log_validation
src/evaluation/monitor.py:115
↓ 2 callersMethodregisterOptionGroup
(e)
lib/tom-select/tom-select.complete.min.js:224
↓ 2 callersFunctionregister_workflow
(spec: WorkflowSpec)
src/evaluation/workflows/registry.py:42
↓ 2 callersMethodremoveActiveItem
(e)
lib/tom-select/tom-select.complete.min.js:196
↓ 2 callersFunctionrender_badges
(d: dict)
scripts/build_spot_check_v2_html.py:57
↓ 2 callersFunctionrender_model_cell
(r: dict)
scripts/build_spot_check_html.py:55
↓ 2 callersMethodresolve_parent
返回父页面的 page_id。 优先级: 1. config.parent_page_id(直接给) 2. data/recorder_state/_parent.json 缓存 3. config.parent_page
src/recorder/notion_client.py:185
↓ 2 callersFunctionrun_one
(label, gold_path, domain, paper_title_hints)
scripts/_eval_gold_samples.py:35
↓ 2 callersFunctionrun_workflow_task
运行单个 L2 工作流任务。
src/evaluation/layers/layer2_workflow.py:81
↓ 2 callersMethodsave_cache
(self)
src/knowledge/semantic_index.py:235
↓ 2 callersFunctionschema_validity_rate
抽取项中 node_type / relation_type 属于合法 Schema 枚举的占比。
src/evaluation/metrics/kg.py:131
↓ 2 callersMethodscroll
(e,t)
lib/tom-select/tom-select.complete.min.js:199
↓ 2 callersMethodscrollToOption
(e,t)
lib/tom-select/tom-select.complete.min.js:197
↓ 2 callersFunctionsearch_arxiv
通过 arXiv API 搜索预印本论文。**推荐作为搜索主路径**(公开 API 无 Key 无限速)。 Phase 4.1 增强: - category 过滤(cs.AI / cs.CL / cs.LG / cs.IR / eess.SP ...) - aut
src/mcp_servers/paper_search.py:498
↓ 2 callersMethodsemantic_search
节点级语义检索。 Phase 2.4 升级:优先用 src/knowledge/semantic_index.py 的 SemanticIndex (sentence-transformers backend);不可用时退化到本类自带的 TF-ID
src/knowledge/graph_store.py:587
↓ 2 callersMethodsetValue
(e,t)
lib/tom-select/tom-select.complete.min.js:191
↓ 2 callersMethodsetupOptions
(e=[],t=[])
lib/tom-select/tom-select.complete.min.js:150
↓ 2 callersMethodspan
计时一段代码并写出 TraceEvent。 span 在记录异常后会重新抛出(ok=False)。 调用方可以用 span.set_result() 覆盖默认的 ok=True。
src/evaluation/tracer.py:73
↓ 2 callersFunctionstdout_assertion_pass
stdout 包含所有期望子串的比例。
src/evaluation/metrics/capability.py:136
↓ 2 callersMethodsummarize_run
(self, run_log: dict)
src/runtime/reflector.py:91
↓ 2 callersMethodsummary
(self)
src/evaluation/monitor.py:73
↓ 2 callersFunctionsy
(t)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersMethodsync
(e=!0)
lib/tom-select/tom-select.complete.min.js:153
↓ 2 callersMethodto_dict
(self)
src/evaluation/schema.py:322
↓ 2 callersMethodto_dict
(self)
scripts/ingest_comms_v1.py:123
↓ 2 callersMethodto_markdown
转为可展示的 Markdown 格式
src/core/multimodal.py:75
↓ 2 callersMethodto_summary
生成抽取结果的摘要
src/knowledge/schema.py:459
↓ 2 callersFunctiontool_sequence_match
实际工具调用序列与期望序列的逐位匹配率。
src/evaluation/metrics/workflow.py:22
↓ 2 callersFunctiontool_success_rate
ok=True 的 tool_call / workflow_step 事件占比。tool_name 给定时只统计该工具。
src/evaluation/metrics/common.py:15
↓ 2 callersFunctiontp
(t)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionuO
(t)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersMethoduncacheValue
(e)
lib/tom-select/tom-select.complete.min.js:297
↓ 2 callersMethodunlock
()
lib/tom-select/tom-select.complete.min.js:292
↓ 2 callersFunctionvO
(t,e)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionvb
(t)
lib/vis-9.1.2/vis-network.min.js:26
↓ 2 callersFunctionverdict
(r: dict)
scripts/probe_vision.py:110
↓ 2 callersFunctionvm
(t,e,i)
lib/vis-9.1.2/vis-network.min.js:26
← previousnext →401–500 of 1,438, ranked by callers