File size: 6,638 Bytes
8a1c7c6
0b2af11
8a1c7c6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
{
  "task_id": "comptox_011",
  "domain": "COMPTOX_CHEMEXPO",
  "autonomy_type": "ordered table",
  "oracle_output_cardinality": 6,
  "instruction": "我想在 ChemExpo 里看看这六个物质,当前到底是更多出现在材料接触相关的用途里,还是制造来源相关的用途里,然后从对应的方向上各选出一个最能代表它的用途分类。请按下面的步骤来查。先把范围限定在这六个物质:Di(2-Ethylhexyl) Phthalate、Melamine、Tris(2-Ethylhexyl) Phosphate、4,5-Dihydro-2-Mercaptoimidazole、Phthalic Acid、Triphenyl Phosphates Tert-Butylated。然后,对每个物质先看五类跟材料接触有关的标签有没有出现过,每类出现了就记 1 分:Indirect additives food contact (10/2018);Europe, Food contact items;Food contact items, Japan;任何带有 Toys and children's products 字样的关键词;任何带有 artificial_saliva 或 artificial_sweat 字样的关键词接着,再看两类跟制造来源有关的标签有没有出现过,每类出现了也记 1 分:drinking_water, Europe, manufacturing, plastic_additive;Europe, plastic_additive。另外,再单独统计以下四个带日期的关键词有没有出现过:Indirect additives food contact (10/2018);OEHHA Proposition 65 (3/2019);WA Children's Safe Product Act (4/2020);OEHHA Proposition 65 (1/2023)。要记录命中了几个,以及命中项里最早和最晚的年月。然后看哪边分数高来决定方向:如果材料接触那边的得分严格大于制造来源那边,就按 Article → Formulation → Occupation 的顺序来找最能代表这个物质的类别;否则就按 Occupation → Article → Formulation 的顺序来找。确定了类别之后,从这一层开始,每一层都只沿着产品数最多的下一级分类往下走,一直走到不能再细分为止;如果同一层有并列,就取 PUC ID 更小的。",
  "start_url": "https://comptox.epa.gov/chemexpo/",
  "output_format": "最后,把所有物质按带日期命中数从高到低排列;命中数相同的,再按入选节点的产品数从高到低排列,输出格式如下:<DTXSID>|<contactHits>|<sourceHits>|<datedHits>|<earliest>|<latest>|<prioritySide>|<PUCkind>|<generalPUCID>|<terminalPUCID>|<level>|<Products>|<PUCDocuments>|<PUCCuratedChemicals>结果之间用英文逗号分隔;如果带日期命中数为 0,则 earliest 和 latest 都写 NA。",
  "oracle_answer": "DTXSID5020607|5|2|4|2018-10|2023-01|contact|Article|309|493|family|1|23|58,DTXSID5020601|4|1|3|2018-10|2023-01|contact|Article|309|309|general|1|5836|1827,DTXSID6020802|4|2|1|2018-10|2018-10|contact|Article|313|313|general|15|1104|675,DTXSID8021484|3|1|1|2018-10|2018-10|contact|Occupation|321|321|general|8|28313|4200,DTXSID0021414|5|0|1|2018-10|2018-10|contact|Article|309|309|general|3|5836|1827,DTXSID20872645|0|1|0|NA|NA|source|Occupation|318|381|family|5|274|107",
  "metadata": {
    "State-Gated Retrieval": [
      "只看题目指定的六个物质,并先在 chemical page 上计算 contactHits、sourceHits 和 datedHits。",
      "prioritySide 必须先由这些历史信号决定,然后再沿对应方向挑选一个最能代表的当前用途分类。",
      "当前路线选择必须按题目给定的 kind 优先级和同层更小 PUC ID 规则继续下钻。"
    ],
    "dependency_type": "Data + Control",
    "intra_chain": true,
    "inter_chain": true,
    "data_dependency": [
      "each chemical page yields contactHits, sourceHits, datedHits, and the earliest/latest dated-hit months from the historical keyword set",
      "after the priority side is chosen, the corresponding current PUC drilldown yields the representative kind, PUC path, product counts, and curated-chemical counts",
      "final rows are produced only after the historical hit scores determine the side first and the current route is then chosen with the side-specific priority and tie-break rules"
    ],
    "control_dependency": [
      "historical contact/source signal scores must take priority over current maximum-Products paths",
      "the priority side must be determined from contactHits, sourceHits, and datedHits rather than from current visibility alone",
      "PUC path tie-breaks are valid only after the correct side has been fixed and the route has been re-drilled under that side's rules"
    ],
    "freeze": {
      "historical_window": "current ChemExpo chemical pages and PUC drilldowns for the six named substances, using the stated historical keyword families and dated hits through 2023-01"
    },
    "answer_type": "multi-row ordered table"
  },
  "rubric": {
    "inclusion_conditions": [
      "只看题目指定的六个物质,并先在 chemical page 上计算 contactHits、sourceHits 和 datedHits。",
      "prioritySide 必须先由这些历史信号决定,然后再沿对应方向挑选一个最能代表的当前用途分类。",
      "当前路线选择必须按题目给定的 kind 优先级和同层更小 PUC ID 规则继续下钻。"
    ],
    "exclusion_conditions": [
      "排除忽略历史信号、直接在 chemical page 上选当前可见 Products 最大路径的结果。",
      "排除未先确定 contact/source 方向就开始做 PUC 路线选择的结果。",
      "排除同层 tie-break 不按更小 PUC ID 规则处理的结果。"
    ],
    "normalization": {
      "field_separator": "|",
      "record_separator": ",",
      "schema": [
        "DTXSID",
        "contactHits",
        "sourceHits",
        "datedHits",
        "earliest",
        "latest",
        "prioritySide",
        "PUCkind",
        "generalPUCID",
        "terminalPUCID",
        "level",
        "Products",
        "PUCDocuments",
        "PUCCuratedChemicals"
      ],
      "dated_hits": "count only the exact dated keyword strings defined in the source rubric",
      "date_format": "YYYY-MM; if datedHits is 0, earliest and latest are both `NA`",
      "prioritySide": "emit `contact` when contactHits > sourceHits; otherwise emit `source`",
      "kind": "the PUCkind field uses the canonical labels Article, Formulation, or Occupation",
      "level": "the level field uses the canonical labels general, family, or type",
      "products_source": "Products comes from the selected terminal chemical-detail node",
      "terminal_statistics": "PUCDocuments and PUCCuratedChemicals come from the selected terminal PUC detail page Additional Statistics block",
      "sorting_or_selection": {
        "primary": "datedHits descending",
        "secondary": "Products descending",
        "tertiary": "DTXSID ascending"
      }
    }
  }
}