情绪分析数据未显示在 csv 文件中

Question

我将数据放入一个 csv 文件（称为“Essential Data_posts”）。在我的主要内容中，我从这个文件中提取了一个特定的列（称为“帖子文本”），以便我可以使用 Google Cloud NLP 分析帖子文本以进行情感实体分析。然后我将此分析放入另一个 csv 文件（称为“SentimentAnalysis”）中。为此，我将与情感实体分析有关的所有信息放入一个数组中（每条信息一个）。

我遇到的问题是，当我执行我的代码时，SentimentAnalysis 文件中除了标题之外什么都没有显示，例如。 “代表名称”。当我请求所有数组的长度时，我发现每个数组的长度都是0，所以它们没有添加信息。

我使用的是 Ubuntu 21.04 和 Google Cloud Natural Language。我在终端中运行这一切，而不是谷歌云平台。我也在使用 Python3 和 emacs 文本编辑器。

from google.cloud import language_v1
import pandas as pd
import csv
import os

#lists we are appending to
representativeName = []
entity = []
salienceScore = []
entitySentimentScore = []
entitySentimentMagnitude = []
metadataNames = []
metadataValues = []
mentionText = []
mentionType = []

def sentiment_entity(postTexts):
    client = language_v1.LanguageServiceClient()
    type_ = language_v1.Document.Type.PLAIN_TEXT
    language = "en"
    document = {"content": post_texts, "type": type_, "language": language}
    encodingType = language_v1.EncodingType.UTF8
    response = client.analyze_entity_sentiment(request = {'document': document, 'encoding type': encodingType})

    #loop through entities returned from the API
    for entity in response.entities:
        representativeName.append(entity.name)
        entity.append(language_v1.Entity.Type(entity.type_).name)
        salienceScore.append(entity.salience)
        entitySentimentScore.append(sentiment.score)
        entitySentimentMagnitude.append(sentiment.magnitude)
    
    #loop over metadata associated with entity 
    for metadata_name, metadata_value in entity.metadata.items():
        metadataNames.append(metadata_name)
        metadataValues.append(metadata_value)

    #loop over the mentions of this entity in the input document
    for mention in entity.mentions:
        mentionText.append(mention.text.content)
        mentionType.append(mention.type_)

#put the lists into the csv file (using pandas)    
data = {
    "Representative Name": representativeName,
    "Entity": entity,
    "Salience Score": salienceScore,
    "Entity Sentiment Score": entitySentimentScore,
    "Entity Sentiment Magnitude": entitySentimentMagnitude,
    "Metadata Name": metadataNames,
    "Metadata Value": metadataValues,
    "Mention Text": mentionText,
    "Mention Type": mentionType  
}

df = pd.DataFrame(data)
df
df.to_csv("SentimentAnalysis.csv", encoding='utf-8', index=False)

def main():
    import argparse

    #read the csv file containing the post text we need to analyze
    filename = open('Essential Data_posts.csv', 'r')

    #create dictreader object
    file = csv.DictReader(filename)

    postTexts = []

    #iterate over each column and append values to list
    for col in file:
    postTexts.append(col['Post Text'])

    parser = arg.parse.ArgumentParser()
    parser.add_argument("--postTexts", type=str, default=postTexts)
    args = parser.parse_args()

    sentiment_entity(args.postTexts)

Answer 1

我尝试运行您的代码，但遇到以下错误：

1. 您没有在sentiment_entity() 中使用传递的参数postTexts，因此这将在document = {"content": post_texts, "type": type_, "language": language} 处出错。
2. 列表不能传递给"content": post_texts，它应该是字符串。见Document reference。
3. 在变量request中，'encoding type'应该是'encoding_type'
4. 局部变量entity 不应与entity = [] 同名。 Python 将尝试在不是列表的局部变量 entity 中附加值。
5. 应该是entity.sentiment.score 和entity.sentiment.magnitude 而不是sentiment.score 和sentiment.magnitude
6. metadata 和 mention 的循环应在循环 for entity in response.entities: 下

我编辑了您的代码并修复了上述错误。在您的main() 中，我包含了一个将列表postTexts 转换为字符串的步骤，以便它可以在您的sentiment_entity() 函数中使用。 metadataNames 和 metadataValues 暂时被评论，因为我没有可以填充这些值的示例。

from google.cloud import language_v1
import pandas as pd
import csv
import os

#lists we are appending to
representativeName = []
entity_arr = []
salienceScore = []
entitySentimentScore = []
entitySentimentMagnitude = []
metadataNames = []
metadataValues = []
mentionText = []
mentionType = []

def listToString(s):
    """ Transform list to string"""
    str1 = " "
    return (str1.join(s))
    
def sentiment_entity(postTexts):
    client = language_v1.LanguageServiceClient()
    type_ = language_v1.Document.Type.PLAIN_TEXT
    language = "en"
    document = {"content": postTexts, "type_": type_, "language": language}
    encodingType = language_v1.EncodingType.UTF8
    response = client.analyze_entity_sentiment(request = {'document': document, 'encoding_type': encodingType})

    #loop through entities returned from the API
    for entity in response.entities:
        representativeName.append(entity.name)
        entity_arr.append(language_v1.Entity.Type(entity.type_).name)
        salienceScore.append(entity.salience)
        entitySentimentScore.append(entity.sentiment.score)
        entitySentimentMagnitude.append(entity.sentiment.magnitude)
        #loop over the mentions of this entity in the input document
        for mention in entity.mentions:
            mentionText.append(mention.text.content)
            mentionType.append(mention.type_)
        #loop over metadata associated with entity
        for metadata_name, metadata_value in entity.metadata.items():
            metadataNames.append(metadata_name)
            metadataValues.append(metadata_value)

    data = {
    "Representative Name": representativeName,
    "Entity": entity_arr,
    "Salience Score": salienceScore,
    "Entity Sentiment Score": entitySentimentScore,
    "Entity Sentiment Magnitude": entitySentimentMagnitude,
    #"Metadata Name": metadataNames,
    #"Metadata Value": metadataValues,
    "Mention Text": mentionText,
    "Mention Type": mentionType
    }

    df = pd.DataFrame(data)
    df.to_csv("SentimentAnalysis.csv", encoding='utf-8', index=False)

def main():
    import argparse

    #read the csv file containing the post text we need to analyze
    filename = open('test.csv', 'r')

    #create dictreader object
    file = csv.DictReader(filename)

    postTexts = []

    #iterate over each column and append values to list
    for col in file:
        postTexts.append(col['Post Text'])
    content = listToString(postTexts) #convert list to string
    print(content)
    sentiment_entity(content)

if __name__ == "__main__":
    main()

---

test.csv：

col_1,Post Text
dummy,Grapes are good.
dummy,Bananas are bad.

运行代码时，我将转换后的列表打印为字符串并生成 SentimentAnalysis.csv：

SentimentAnalysis.csv：

Representative Name,Entity,Salience Score,Entity Sentiment Score,Entity Sentiment Magnitude,Mention Text,Mention Type
Grapes,OTHER,0.8335162997245789,0.800000011920929,0.800000011920929,Grapes,2
Bananas,OTHER,0.16648370027542114,-0.699999988079071,0.699999988079071,Bananas,2