NLP入门spaCy配置_spacy库安装与配置

​
我用的是python， 打开终端，执行以下代码（参考：https://spacy.io/usage）：

pip install -U pip setuptools wheel
pip install -U spacy
python -m spacy download en_core_web_sm
1
2
3

其中第二行pip install -U spacy时,出现提示让我弄C++环境，于是按照提示下载了vs_buildtools，然后安装完之后打开把C++相关的右侧列表中的几乎都点了对勾，然后安装，大概需要半小时。（因为我不懂所以干脆全部安装）

然后再继续上面的第二行代码就成功了

然鹅，第三行代码python -m spacy download en_core_web_sm运行又有问题了，于是只好手动下载en_core_web_sm。
先pip list查看spacy的版本，我的是3.7.5。然后去这个网站 https://github.com/explosion/spacy-models/releases 找到页面最下方的en_core_web_trf-3.7.3.tar.gz下载，然后pip 安装试试。

成功，然后运行下面的程序：

# chatgpt generated 

import spacy

# Load spaCy's pre-trained model
nlp = spacy.load("en_core_web_trf")
 
# Transcribed sentence from Vosk
sentence = "Please pick up the red box and put it on the desk in front of me."

# Process the sentence using spaCy
doc = nlp(sentence)

# Extracting actions (verbs) and their objects
actions = []
objects = []
places = []

for token in doc:
    if token.pos_ == "VERB":
        actions.append(token.text)
        for child in token.children:
            if child.dep_ in ("dobj", "attr"):  # Direct object or attribute
                objects.append(child.text)
                # Get the full noun phrase
                noun_phrase = ' '.join([child.text] + [grandchild.text for grandchild in child.children if grandchild.dep_ in ("amod", "compound")])
                objects.append(noun_phrase)
            if child.dep_ in ("prep"):  # Prepositional phrase
                place_phrase = ' '.join([child.text] + [grandchild.text for grandchild in child.children])
                places.append(place_phrase)

# Extracting target locations
for chunk in doc.noun_chunks:
    if any(substring in chunk.text for substring in ["desk", "me"]):
        places.append(chunk.text)

# Remove duplicates
objects = list(set(objects))
places = list(set(places))

# Display results
print("Actions:", actions)
print("Targeted Objects:", objects)
print("Targeted Places:", places)

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

​运行结果如下：

# 使用trf的模型时(accuracy)：
[Running] set PYTHONIOENCODING=utf8 && python -u "d:\Software\Python\pyCode\voice\spaCy\test.py"
Actions: ['pick', 'put']
Targeted Objects: ['box', 'box red', 'it']
Targeted Places: ['on desk', 'me', 'the desk']

[Done] exited with code=0 in 5.135 seconds


# 使用sm的模型时(efficiency)：
[Running] set PYTHONIOENCODING=utf8 && python -u "d:\Software\Python\pyCode\voice\spaCy\test.py"
Actions: ['pick', 'put']
Targeted Objects: ['box', 'it', 'box red']
Targeted Places: ['the desk', 'on desk', 'me']

[Done] exited with code=0 in 2.182 seconds

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

这两个模型，如果手动安装的话，可以在 https://spacy.io/models/en#en_core_web_trf 这里下载。