ddi_corpus

Name: ddi_corpus
Author: QSong-github

132stars4forks

How to install ddi_corpus?

qsong-github/drugclaw/ddi_corpus

$npx -y skills add QSong-github/DrugClaw --skill ddi_corpus

Installs into the current project.

Summary

| Field | Value | |-------|-------| | Resource | DDI Corpus 2013 | | Category | Drug-centric / Drug NLP & Text Mining | | Source | GitHub | | Paper | [Herrero-Zazo et al., 2013](https://www.sciencedirect.com/science/article/

Files · 1

View on GitHub

SKILL.md

1	# DDI Corpus 2013 – Drug-Drug Interaction Query Skill
2
3	## Overview
4
5	\| Field \| Value \|
6	\|-------\|-------\|
7	\| Resource \| DDI Corpus 2013 \|
8	\| Category \| Drug-centric / Drug NLP & Text Mining \|
9	\| Source \| [GitHub](https://github.com/isegura/DDICorpus) \|
10	\| Paper \| [Herrero-Zazo et al., 2013](https://www.sciencedirect.com/science/article/pii/S1532046413001123) \|
11	\| Corpus Size \| ~2,740 unique entities, ~5,000 annotated DDI pairs \|
12	\| Sources \| DrugBank descriptions + MEDLINE abstracts \|
13
14	The DDI Corpus 2013 is the standard benchmark for drug-drug interaction (DDI) extraction from biomedical text. Each XML file contains sentences with annotated drug entities and pairwise DDI labels.
15
16	DDI Types:
17	- `mechanism` – pharmacokinetic mechanism described (e.g., altered absorption/metabolism)
18	- `effect` – clinical effect of the interaction (e.g., increased bleeding risk)
19	- `advise` – recommendation or warning about co-administration
20	- `int` – stated interaction without further detail
21
22	Entity Types: `drug`, `group`, `brand`, `drug_n` (active substance not approved for human use)
23
24	## Setup
25
26	1. Download & extract (one-time):
27
28	```bash
29	git clone https://github.com/isegura/DDICorpus.git
30	cd DDICorpus
31	unzip DDICorpus-2013.zip
32	```
33
34	2. Set the corpus path in `30_DDI_Corpus_2013.py`:
35
36	```python
37	CORPUS_ROOT = "/path/to/DDICorpus-master" # contains DDICorpus/Train/ and DDICorpus/Test/
38	```
39
40	Or pass `--root` at runtime or set env var `DDI_CORPUS_ROOT`.
41
42	## Usage
43
44	### Python API
45
46	```python
47	from 30_DDI_Corpus_2013 import query_entities, list_all_entities, corpus_stats
48
49	# Query a single drug
50	result = query_entities("aspirin")
51
52	# Query multiple drugs at once
53	result = query_entities(["warfarin", "metformin", "digoxin"])
54
55	# List all entity names in the corpus
56	names = list_all_entities()
57
58	# Get corpus-level statistics
59	stats = corpus_stats()
60	```
61
62	### CLI (直接运行)
63
64	```bash
65	python 30_DDI_Corpus_2013.py
66	```
67
68	直接运行即输出 demo 结果（corpus 统计 → 单实体查询 → 批量查询 → 未找到示例）。
69	修改 `__main__` 块中的实体名即可自定义查询。
70
71	## Output Format
72
73	`query_entities` returns a JSON string. Each element:
74
75	```json
76	{
77	"query": "aspirin",
78	"found": true,
79	"canonical_names": ["ASPIRIN", "Aspirin", "aspirin"],
80	"entity_types": ["brand", "drug"],
81	"total_interactions": 65,
82	"interactions": [
83	{
84	"partner": "ketoprofen",
85	"ddi_type": "mechanism",
86	"sentence": "concurrent administration of aspirin decreased ketoprofen protein binding...",
87	"source": "Train/DrugBank"
88	}
89	],
90	"example_sentences": ["..."]
91	}
92	```
93
94	If an entity is not found: `{"query": "xyz", "found": false}`.
95
96	## Parameters
97
98	\| Parameter \| Default \| Description \|
99	\|-----------\|---------\|-------------\|
100	\| `entities` \| (required) \| `str` or `list[str]` — drug names to look up (case-insensitive) \|
101	\| `corpus_root` \| `CORPUS_ROOT` \| Path to the extracted DDICorpus-master directory \|
102	\| `max_interactions` \| `20` \| Maximum interaction records returned per entity \|
103	\| `max_sentences` \| `5` \| Maximum example sentences returned per entity \|
104
105	## Dependencies
106
107	Python 3.10+ standard library only (`xml.etree.ElementTree`, `json`, `os`, `collections`). No third-party packages required.

People who liked this also viewed

6 picks

Type

ddi_corpus

132stars4forks

View on GitHub

How to install ddi_corpus?

qsong-github/drugclaw/ddi_corpus

$npx -y skills add QSong-github/DrugClaw --skill ddi_corpus

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# DDI Corpus 2013 – Drug-Drug Interaction Query Skill
2
3	## Overview
4
5	\| Field \| Value \|
6	\|-------\|-------\|
7	\| Resource \| DDI Corpus 2013 \|
8	\| Category \| Drug-centric / Drug NLP & Text Mining \|
9	\| Source \| [GitHub](https://github.com/isegura/DDICorpus) \|
10	\| Paper \| [Herrero-Zazo et al., 2013](https://www.sciencedirect.com/science/article/pii/S1532046413001123) \|
11	\| Corpus Size \| ~2,740 unique entities, ~5,000 annotated DDI pairs \|
12	\| Sources \| DrugBank descriptions + MEDLINE abstracts \|
13
14	The DDI Corpus 2013 is the standard benchmark for drug-drug interaction (DDI) extraction from biomedical text. Each XML file contains sentences with annotated drug entities and pairwise DDI labels.
15
16	DDI Types:
17	- `mechanism` – pharmacokinetic mechanism described (e.g., altered absorption/metabolism)
18	- `effect` – clinical effect of the interaction (e.g., increased bleeding risk)
19	- `advise` – recommendation or warning about co-administration
20	- `int` – stated interaction without further detail
21
22	Entity Types: `drug`, `group`, `brand`, `drug_n` (active substance not approved for human use)
23
24	## Setup
25
26	1. Download & extract (one-time):
27
28	```bash
29	git clone https://github.com/isegura/DDICorpus.git
30	cd DDICorpus
31	unzip DDICorpus-2013.zip
32	```
33
34	2. Set the corpus path in `30_DDI_Corpus_2013.py`:
35
36	```python
37	CORPUS_ROOT = "/path/to/DDICorpus-master" # contains DDICorpus/Train/ and DDICorpus/Test/
38	```
39
40	Or pass `--root` at runtime or set env var `DDI_CORPUS_ROOT`.
41
42	## Usage
43
44	### Python API
45
46	```python
47	from 30_DDI_Corpus_2013 import query_entities, list_all_entities, corpus_stats
48
49	# Query a single drug
50	result = query_entities("aspirin")
51
52	# Query multiple drugs at once
53	result = query_entities(["warfarin", "metformin", "digoxin"])
54
55	# List all entity names in the corpus
56	names = list_all_entities()
57
58	# Get corpus-level statistics
59	stats = corpus_stats()
60	```
61
62	### CLI (直接运行)
63
64	```bash
65	python 30_DDI_Corpus_2013.py
66	```
67
68	直接运行即输出 demo 结果（corpus 统计 → 单实体查询 → 批量查询 → 未找到示例）。
69	修改 `__main__` 块中的实体名即可自定义查询。
70
71	## Output Format
72
73	`query_entities` returns a JSON string. Each element:
74
75	```json
76	{
77	"query": "aspirin",
78	"found": true,
79	"canonical_names": ["ASPIRIN", "Aspirin", "aspirin"],
80	"entity_types": ["brand", "drug"],
81	"total_interactions": 65,
82	"interactions": [
83	{
84	"partner": "ketoprofen",
85	"ddi_type": "mechanism",
86	"sentence": "concurrent administration of aspirin decreased ketoprofen protein binding...",
87	"source": "Train/DrugBank"
88	}
89	],
90	"example_sentences": ["..."]
91	}
92	```
93
94	If an entity is not found: `{"query": "xyz", "found": false}`.
95
96	## Parameters
97
98	\| Parameter \| Default \| Description \|
99	\|-----------\|---------\|-------------\|
100	\| `entities` \| (required) \| `str` or `list[str]` — drug names to look up (case-insensitive) \|
101	\| `corpus_root` \| `CORPUS_ROOT` \| Path to the extracted DDICorpus-master directory \|
102	\| `max_interactions` \| `20` \| Maximum interaction records returned per entity \|
103	\| `max_sentences` \| `5` \| Maximum example sentences returned per entity \|
104
105	## Dependencies
106
107	Python 3.10+ standard library only (`xml.etree.ElementTree`, `json`, `os`, `collections`). No third-party packages required.

People who liked this also viewed

6 picks

Type