使用gosentiwordnet进行情感分析

gosentiwordnet是一个基于SentiWordNet词典的Golang情感分析库。SentiWordNet是一个基于WordNet的情感词典，为每个词条分配了正面、负面和中性情感分数。

安装

首先安装gosentiwordnet库：

go get github.com/dinopuguh/gosentiwordnet

基本用法

初始化情感分析器

package main

import (
	"fmt"
	"github.com/dinopuguh/gosentiwordnet"
)

func main() {
	// 初始化情感分析器
	swn := gosentiwordnet.New()

	// 分析单个词的情感
	score := swn.GetScore("happy")
	fmt.Printf("happy: Pos=%.3f, Neg=%.3f, Obj=%.3f\n", 
		score.Positive, score.Negative, score.Objective)
	
	// 分析句子情感
	sentence := "I love this beautiful day but hate the traffic"
	sentenceScore := swn.AnalyzeSentence(sentence)
	fmt.Printf("Sentence: %s\nPos=%.3f, Neg=%.3f, Obj=%.3f\n",
		sentence,
		sentenceScore.Positive,
		sentenceScore.Negative,
		sentenceScore.Objective)
}

输出结果示例

happy: Pos=0.750, Neg=0.000, Obj=0.250
Sentence: I love this beautiful day but hate the traffic
Pos=0.600, Neg=0.300, Obj=0.100

高级功能

1. 自定义词典路径

默认情况下，gosentiwordnet使用内置的SentiWordNet数据。你也可以提供自定义路径：

swn := gosentiwordnet.NewWithPath("/path/to/SentiWordNet.txt")

2. 分析文本情感倾向

func analyzeTextSentiment(text string) string {
	swn := gosentiwordnet.New()
	score := swn.AnalyzeSentence(text)
	
	if score.Positive > score.Negative {
		return "Positive"
	} else if score.Negative > score.Positive {
		return "Negative"
	}
	return "Neutral"
}

func main() {
	text := "This movie was terrible and boring"
	fmt.Println(analyzeTextSentiment(text)) // 输出: Negative
}

3. 批量分析文本

func batchAnalyze(texts []string) []gosentiwordnet.Sentiment {
	swn := gosentiwordnet.New()
	results := make([]gosentiwordnet.Sentiment, len(texts))
	
	for i, text := range texts {
		results[i] = swn.AnalyzeSentence(text)
	}
	
	return results
}

func main() {
	reviews := []string{
		"Excellent product, highly recommended",
		"Poor quality, would not buy again",
		"It's okay, nothing special",
	}
	
	scores := batchAnalyze(reviews)
	for i, score := range scores {
		fmt.Printf("Review %d: Pos=%.2f, Neg=%.2f\n", i+1, score.Positive, score.Negative)
	}
}

性能优化

对于大量文本分析，可以预加载词典：

swn := gosentiwordnet.New()
swn.Preload() // 预加载词典到内存

注意事项

1. SentiWordNet主要针对英语单词，对非英语文本效果有限
2. 情感分析结果仅供参考，不能完全依赖
3. 对于短文本，单个强烈情感词可能主导整个文本的情感倾向
4. 不支持上下文理解和讽刺检测

完整示例

package main

import (
	"fmt"
	"github.com/dinopuguh/gosentiwordnet"
)

type SentimentResult struct {
	Text      string
	Sentiment string
	Score     gosentiwordnet.Sentiment
}

func analyzeReviews(reviews []string) []SentimentResult {
	swn := gosentiwordnet.New()
	swn.Preload()
	
	results := make([]SentimentResult, len(reviews))
	
	for i, review := range reviews {
		score := swn.AnalyzeSentence(review)
		sentiment := "Neutral"
		
		if score.Positive > score.Negative + 0.1 { // 添加阈值避免微小差异
			sentiment = "Positive"
		} else if score.Negative > score.Positive + 0.1 {
			sentiment = "Negative"
		}
		
		results[i] = SentimentResult{
			Text:      review,
			Sentiment: sentiment,
			Score:     score,
		}
	}
	
	return results
}

func main() {
	productReviews := []string{
		"Absolutely love this product! Works perfectly.",
		"Waste of money. Broke after two days.",
		"It's fine, does what it's supposed to do.",
		"Horrible customer service and bad quality.",
		"Best purchase I've made this year!",
	}
	
	results := analyzeReviews(productReviews)
	
	for _, result := range results {
		fmt.Printf("Review: %s\n", result.Text)
		fmt.Printf("Sentiment: %s (Pos=%.2f, Neg=%.2f)\n\n", 
			result.Sentiment, 
			result.Score.Positive, 
			result.Score.Negative)
	}
}

gosentiwordnet提供了一个简单有效的方法来执行基于词典的情感分析。对于更复杂的需求，可能需要结合机器学习方法或使用更高级的自然语言处理技术。