基于python-Django的微博舆情情感分析

闲来无事，整理了一下之前做过的项目，这个项目为基于python开发的微博舆情情感分析

1.使用的框架及包

jieba
Django
snownlp
sqlite
pandas
tqdm
scikit_learn
pyecharts
scrapy
query
Pillow
pymongo

2.系统功能及实现方法

2.1登录界面

界面展示：

 这个系统登陆界面有登录和注册功能，对应的实现代码分别为：

注册：

def register(request):
    if request.method == "POST":
        name, tel, pwd = request.POST.get('name'), request.POST.get('tel'), request.POST.get('pwd')
        print(name, tel, pwd)
        if User.objects.filter(tel=tel):
            msg = "你已经有账号了，请登录"
        else:
            User.objects.create(name=name, tel=tel, password=pwd)
            msg = "注册成功，请登录！"
        return render(request, 'login.html', locals())
    else:
        msg = ""
        return render(request, 'register.html', locals())

登录：

def login(request):
    if request.method == "POST":
        tel, pwd = request.POST.get('tel'), request.POST.get('pwd')
        if User.objects.filter(tel=tel, password=pwd):

            obj = redirect('/') #将用户重定向到网站的首页。
            obj.set_cookie('uid', User.objects.filter(tel=tel, password=pwd)[0].id, max_age=60 * 60 * 24)
            return obj
        else:
            msg = "用户信息错误，请重新输入！！"
            return render(request, 'login.html', locals())
    else:
        return render(request, 'login.html', locals())

 2.2功能界面-话题选择

 

可以选择当前最热话题，通过下拉菜单进行选择。

def index(request):
    # 话题列表
    topic_raw = [item.topic for item in WeiBo.objects.all() if item.topic]
    topic_list = []
    for item in topic_raw:
        topic_list.extend(item.split(','))
    topic_list = list(set(topic_list))
    # yon用户信息
    uid = int(request.COOKIES.get('uid', -1))
    if uid != -1:
        username = User.objects.filter(id=uid)[0].name
    # 得到话题
    if 'key' not in request.GET:
        key = topic_list[0]
        raw_data = WeiBo.objects.all()
    else:
        key = request.GET.get('key')
        raw_data = WeiBo.objects.filter(topic__contains=key)
    # 分页
    if 'page' not in request.GET:
        page = 1
    else:
        page = int(request.GET.get('page'))
    data_list = raw_data[(page - 1) * 20: page * 20]

    return render(request, 'index.html', locals())

2.3 正-负向情感分类可视化界面

ef fenlei(request):
    from snownlp import SnowNLP
  
    # print(s.sentiments)

    for item in tqdm(WeiBo.objects.all()):
        emotion = '正向' if SnowNLP(item.content).sentiments >0.45 else '负向'
        # 对每个对象的content字段进行情感分析（这里使用了SnowNLP库），根据情感分析结果设置情感字段的值为'正向'或者'负向'
        WeiBo.objects.filter(id=item.id).update(emotion=emotion)
    return JsonResponse({'status':1,'msg':'操作成功'} )

2.4 Kmeans聚类分析

from sklearn.feature_extraction.text import CountVectorizer, TfidfTransformer
def kmeansPlot(request):
    uid = int(request.COOKIES.get('uid', -1))
    if uid != -1:
        username = User.objects.filter(id=uid)[0].name

    # 聚类个数
    if 'num' in request.GET:
        num = int(request.GET.get('num'))
    else:
        num = 2
    ### 训练
    # 清洗文本
    clean_data = [item.content for item in WeiBo.objects.all()]
    clean_data = [clearTxt(item) for item in clean_data]
    clean_data = [sent2word(item) for item in clean_data]

    # 该类会将文本中的词语转换为词频矩阵，矩阵元素a[i][j] 表示j词在i类文本下的词频
    vectorizer = CountVectorizer(max_features=20000)
   
    tfidf_matrix = X_tfidf.toarray()

    # 聚成5类
    from sklearn.cluster import KMeans
    clf = KMeans(n_clusters=num)
    result_list = clf.fit(tfidf_matrix)
    result_list = list(clf.predict(tfidf_matrix))

    
 

    div_id_list = [f'container{i + 1}' for i in range(num)]

    data_list = []
    for label, name in enumerate(div_id_list):
        tmp = {'id': name, 'data': [], 'title': f'第{label + 1}类'}
        # 汇总
        tmp_text_list = ''
        for la, text in zip(result_list, clean_data):
            if la == label:
                tmp_text_list += ' ' + text
        tmp_text_list = [item for item in tmp_text_list.split(' ') if item.strip() != ' ']

        # 得到前30
        rank_Data = [
            {
                'value': tmp_text_list.count(item),
                'name': item
            }
            for item in set(tmp_text_list)
        ]
        rank_Data = sorted(rank_Data, key=lambda x: x['value'], reverse=True)[:100]

        tmp['data'] = rank_Data
        data_list.append(tmp)

    return render(request, 'kmeansPlot.html', locals())