[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"cms-guide-articles":3},{"articleList":4},[5,19,28,38,48,57,66,76,85,94,103,114,127,137],{"id":6,"title":7,"type":8,"index":9,"lang":10,"category":11,"tags":12,"coverImages":14,"slug":16,"summary":17,"updatedAt":18,"createdAt":18},"6a5c3391bed7be6c9f36a7b6","DeepSentiV2 情感分析教程：按语境判断整句倾向","admin_cms",0,"zh-cn","guide",[13],"文本分析基础教程",[15],"https:\u002F\u002Flatte-api-uploader.oss-cn-chengdu.aliyuncs.com\u002Fanna\u002Fadmin-cms\u002Fdraft\u002Fcover\u002Fd8e625b90ed7da7551ad70ebdc4e7555.png","deepsentiv2-qing-gan-fen-xi-jiao-cheng-an-yu-jing-pan-duan-zheng-ju-qing-xiang","DeepSentiV2 用整句语义而不是单个褒贬词判断情感。每条有效文本会得到积极、中性或消极标签、0 到 1 的得分，以及最多五个关键词。工具还提供七种行业文本类型，让同一句表达放回更贴近业务的语境中分析。下面从文本粒度、参数选择、报告阅读和两组实测样本说明具体用法。","2026-07-19T02:16:49.142Z",{"id":20,"title":21,"type":8,"index":9,"lang":10,"category":11,"tags":22,"coverImages":23,"slug":25,"summary":26,"updatedAt":27,"createdAt":27},"6a362f7563012f7eb231ace4","传统情感分析工具使用教程：对比两种词典方法，补充自己的情感词表",[13],[24],"https:\u002F\u002Flatte-api-uploader.oss-cn-chengdu.aliyuncs.com\u002Fanna\u002Fadmin-cms\u002Fdraft\u002Fcover\u002Fb56c36027f92a5605299e467a5f16ca4.png","chuan-tong-qing-gan-fen-xi-gong-ju-shi-yong-jiao-cheng-dui-bi-liang-zhong-ci-dian-fang-fa-bu-chong-zi-ji-de-qing-gan-ci-biao","传统情感分析基于 PySenti 和 CnText 两个开源词典方法做二次开发，底层词典融入了多年实际项目中积累的行业情感词。它适合做评论、问卷开放题、舆情短文本和访谈分段的初步正负中标注。这个工具的重点不是自动理解复杂语境，而是把两种词典方法放在同一份报告里对比，并允许用户自己补充积极词和消极词。你可以用它查看 PySenti 在默认词典和自建词典下的变化，同时把 CnText 作为另一套内置情绪词典口径参考，再把分歧样本交给人工复核。","2026-06-20T06:13:09.446Z",{"id":29,"title":30,"type":8,"index":9,"lang":10,"category":11,"tags":31,"coverImages":32,"slug":34,"summary":35,"updatedAt":36,"createdAt":37},"6a362dd763012f7eb231a001","文本矩阵分析工具使用教程：给多份材料定核心词、看词与词怎么抱团",[13],[33],"https:\u002F\u002Flatte-api-uploader.oss-cn-chengdu.aliyuncs.com\u002Fanna\u002Fadmin-cms\u002Fdraft\u002Fcover\u002F96408dcc2e784ade4e43ba6d85ab5d8e.png","wen-ben-ju-zhen-fen-xi-gong-ju-shi-yong-jiao-cheng-gei-duo-fen-cai-liao-ding-he-xin-ci-kan-ci-yu-ci-zen-mo-bao-tuan","文本矩阵分析适合回答一个问题：这批材料里哪些词最值得关注，它们之间又是怎么抱团的。你上传多份 TXT 或 CSV，系统会先把文本切成一段段「文档段」，统计词频并做 TF-IDF 加权，挑出在这批材料里区分度最高的核心词；再看这些词在同一段里谁和谁反复结伴出现，连成共现矩阵和关系网络。它不替你归纳主题，而是给你一份按重要度排好的词表、一张谁与谁结伴的热力图，以及每个词对回原文核对的证据，帮你决定从哪里开始读、从哪里开始编码。","2026-06-20T06:08:24.387Z","2026-06-20T06:06:15.315Z",{"id":39,"title":40,"type":8,"index":9,"lang":10,"category":11,"tags":41,"coverImages":42,"slug":44,"summary":45,"updatedAt":46,"createdAt":47},"6a2e65110436dbaa6af8b71d","依存句法分析工具使用教程：可视化句子骨架，量化句式复杂度",[13],[43],"https:\u002F\u002Flatte-api-uploader.oss-cn-chengdu.aliyuncs.com\u002Fanna\u002Fadmin-cms\u002Fdraft\u002Fcover\u002F1d5f4075e679558f247510a3c6ccee5b.png","yi-cun-ju-fa-fen-xi-gong-ju-shi-yong-jiao-cheng-ke-shi-hua-ju-zi-gu-jia-liang-hua-ju-shi-fu-za-du","依存句法分析适合回答「这句话的结构是什么样的、哪里让读者读不懂」。编辑处理长难句时需要定位主干，老师讲解句式时需要可视化结构，研究者对比句式复杂度时需要量化指标——这些场景的共同点是：不能只靠语感判断句子结构，还需要看得见的骨架和可比较的数字。这个工具会逐句标出主语、谓语、宾语和修饰成分之间的依赖关系，用可视化依存关系图呈现句子结构，并给出句法复杂度评分。","2026-06-20T06:08:00.294Z","2026-06-14T08:23:45.771Z",{"id":49,"title":50,"type":8,"index":9,"lang":10,"category":11,"tags":51,"coverImages":52,"slug":54,"summary":55,"updatedAt":56,"createdAt":56},"6a2cc703d997a5f0270d666c","词语共现分析工具使用教程：用统计指标判断词对搭配是否可靠",[13],[53],"https:\u002F\u002Flatte-api-uploader.oss-cn-chengdu.aliyuncs.com\u002Fanna\u002Fadmin-cms\u002Fdraft\u002Fcover\u002Ffafae9ffc376439c61a91abd8a053a50.png","ci-yu-gong-xian-fen-xi-gong-ju-shi-yong-jiao-cheng-yong-tong-ji-zhi-biao-pan-duan-ci-dui-da-pei-shi-fou-ke-kao","词语共现分析适合回答「哪些词经常一起出现，这种搭配是否稳定」。做舆情分析时需要梳理话语口径，做访谈研究时需要提取概念关联，做知识图谱时需要准备搭配数据——这些场景的共同点是：不能只看单个词的频率，还要看词和词之间的关系。这个工具会在指定窗口范围内扫描词对，用三种统计指标判断搭配是否超出随机水平，帮你从文本里提取可靠的词语搭配结构。","2026-06-13T02:57:07.768Z",{"id":58,"title":59,"type":8,"index":9,"lang":10,"category":11,"tags":60,"coverImages":61,"slug":63,"summary":64,"updatedAt":65,"createdAt":65},"6a2ac01691a276b608e30952","文体风格指纹工具使用教程：把写作习惯拆成可对比的风格画像",[13],[62],"https:\u002F\u002Flatte-api-uploader.oss-cn-chengdu.aliyuncs.com\u002Fanna\u002Fadmin-cms\u002Fdraft\u002Fcover\u002Faf821f5e176a37198b0520ca02c6a7d8.png","wen-ti-feng-ge-zhi-wen-gong-ju-shi-yong-jiao-cheng-ba-xie-zuo-xi-guan-chai-cheng-ke-dui-bi-de-feng-ge-hua-xiang","文体风格指纹适合把模糊的写作读感拆成具体指标。老师觉得作文有点散、编辑觉得改稿没改到位、审稿人觉得论证不够连贯、运营觉得通稿语气不统一——这些判断往往是对的，但落到修改时还需要更具体的依据。这个工具会从句长、句型、词汇结构、词性占比、标点密度和人称代词使用率等维度生成写作画像，让每个角色的读感都能落到可修改的指标上。","2026-06-11T14:03:02.269Z",{"id":67,"title":68,"type":8,"index":9,"lang":10,"category":11,"tags":69,"coverImages":70,"slug":72,"summary":73,"updatedAt":74,"createdAt":75},"6a2ab66d91a276b608e2961e","命名实体识别工具使用教程：从文本中提取人名、地名、机构和时间线索",[13],[71],"https:\u002F\u002Flatte-api-uploader.oss-cn-chengdu.aliyuncs.com\u002Fanna\u002Fadmin-cms\u002Fdraft\u002Fcover\u002Fdd9c77ccbe5a3e27abd25f53df74c807.png","ming-ming-shi-ti-shi-bie-gong-ju-shi-yong-jiao-cheng-cong-wen-ben-zhong-ti-qu-ren-ming-di-ming-ji-gou-he-shi-jian-xian-suo","命名实体识别适合在正式分析前整理对象清单。它会从 TXT 或 CSV 文本中提取人名、地名、机构名、时间等专名线索，并保留实体所在原句、实体类型、开始位置和结束位置。报告页会同时展示规则识别和语义识别两种口径，上方先给出整体解读，下方用高频实体和句子级对照表帮助你回到原文复核。这个功能不替你判断人物关系或事件因果，它的价值在于把散在文本里的对象先整理出来。","2026-06-11T13:41:43.920Z","2026-06-11T13:21:49.346Z",{"id":77,"title":78,"type":8,"index":9,"lang":10,"category":11,"tags":79,"coverImages":80,"slug":82,"summary":83,"updatedAt":84,"createdAt":84},"6a27de380d824b2ecc5e90bc","高频词提取工具使用教程：统计词频和固定词组，找出文本中反复强调的核心提法",[13],[81],"https:\u002F\u002Flatte-api-uploader.oss-cn-chengdu.aliyuncs.com\u002Fanna\u002Fadmin-cms\u002Fdraft\u002Fcover\u002F4a58ea32ddd52f888d77aef4992a153e.png","gao-pin-ci-ti-qu-gong-ju-shi-yong-jiao-cheng-tong-ji-ci-pin-he-gu-ding-ci-zu-zhao-chu-wen-ben-zhong-fan-fu-qiang-tiao-de-he-xin-ti-fa","写报告、做分析的时候，你可能想知道一篇文章里哪些词被反复提到、哪些说法经常连在一起出现。靠人工通读很难精确量化，靠AI读也不行——容易有幻觉，遗漏或编造根本不存在的提法。高频词提取解决的就是这个问题：统计单个词的出现次数，也可以在你勾选两个词、三个词、四个词组合后，统计相邻词组成的固定词组。报告会自动给出关键发现，配上词云、面积图、Top 20 清单和「核心词 × 长组合」对比表；多文件时还会展示跨文档共有词和单篇独有词。","2026-06-09T09:34:48.273Z",{"id":86,"title":87,"type":8,"index":9,"lang":10,"category":11,"tags":88,"coverImages":89,"slug":91,"summary":92,"updatedAt":93,"createdAt":93},"6a2539fdf98a56c7bb098492","关键词抽取工具使用教程：TF-IDF 和 TextRank 双算法交叉验证，提取文本核心关键词",[13],[90],"https:\u002F\u002Flatte-api-uploader.oss-cn-chengdu.aliyuncs.com\u002Fanna\u002Fadmin-cms\u002Fdraft\u002Fcover\u002F951fedc2c5d4ad704878fee5d743fbfa.png","guan-jian-ci-chou-qu-gong-ju-shi-yong-jiao-cheng-tf-idf-he-textrank-shuang-suan-fa-jiao-cha-yan-zheng-ti-qu-wen-ben-he-xin-guan-jian-ci","一篇文章里哪些词最能代表它的主题？单靠一种算法可能会有偏差。关键词抽取同时运行 TF-IDF 和 TextRank 两种方法，TF-IDF 看稀有度，TextRank 看关联强度，再自动计算排名相关系数，帮你在交叉验证中找到最可靠的关键词。","2026-06-07T09:29:33.876Z",{"id":95,"title":96,"type":8,"index":9,"lang":10,"category":11,"tags":97,"coverImages":98,"slug":100,"summary":101,"updatedAt":102,"createdAt":102},"6a25316df98a56c7bb092d91","词性标注工具使用教程：统计词类分布，分析文本语法特征",[13],[99],"https:\u002F\u002Flatte-api-uploader.oss-cn-chengdu.aliyuncs.com\u002Fanna\u002Fadmin-cms\u002Fdraft\u002Fcover\u002F0c7c2a78832a0f13ebcaf924294e129c.png","ci-xing-biao-zhu-gong-ju-shi-yong-jiao-cheng-tong-ji-ci-lei-fen-bu-fen-xi-wen-ben-yu-fa-te-zheng","文本中的名词、动词、形容词比例，会影响我们对文体风格的判断。词性标注会把每个词归入语法类别，再汇总为分布统计、相邻词性转移矩阵和基准语料对比。中文文本同时跑两套标注引擎，结果可以相互参照，帮助发现分词路径和语法角色判断上的差异；英文走句法级路径，标签集更贴近跨语言语法体系。","2026-06-07T08:53:01.366Z",{"id":104,"title":105,"type":8,"index":106,"lang":10,"category":11,"tags":107,"coverImages":108,"slug":110,"summary":111,"updatedAt":112,"createdAt":113},"69edc5bc6acb5b29015a83fe","文本质量评估器：从字符构成到任务适用性，给语料做一次全面体检",1,[13],[109],"https:\u002F\u002Flatte-api-uploader.oss-cn-chengdu.aliyuncs.com\u002Fanna\u002Fadmin-cms\u002F69edc5bc6acb5b29015a83fe\u002Fcover\u002F5322f90bad088c64b08c5a8ea1717dcc.png","wen-ben-zhi-liang-ping-gu-qi-cong-zi-fu-gou-cheng-dao-ren-wu-shi-yong-xing-gei-yu-liao-zuo-yi-ci-quan-mian-ti-jian","你手头有一份文本，想拿去做主题建模或者情感分析，但不确定质量够不够。符号太多会不会让分词崩？重复行太多会不会让主题模型全是噪声？文本质量评估器帮你回答这些问题。它用纯统计方法（不调用大模型）从字符构成、词汇丰富度、重复率、句长分布、信息熵等多个维度给文本做一次量化体检，输出一份 0-100 的综合得分和六种 NLP 任务的适用性判定。","2026-06-07T08:10:55.590Z","2026-04-26T07:58:52.810Z",{"id":115,"title":116,"type":8,"index":117,"lang":10,"category":11,"tags":118,"coverImages":121,"slug":123,"summary":124,"updatedAt":125,"createdAt":126},"6a251d4ff98a56c7bb085fa3","中文文本规范化工具使用教程：繁简转换、标点统一、数字转正文、拼音转写",2,[13,119,120],"文本规范化","文本清洗",[122],"https:\u002F\u002Flatte-api-uploader.oss-cn-chengdu.aliyuncs.com\u002Fanna\u002Fadmin-cms\u002Fdraft\u002Fcover\u002F633e774a97230d0e2461de209d01dab7.png","zhong-wen-wen-ben-gui-fan-hua-gong-ju-shi-yong-jiao-cheng-fan-jian-zhuan-huan-biao-dian-tong-yi-shu-zi-zhuan-zheng-wen-pin-yin-zhuan-xie","不同来源的中文文本，繁简体、标点符号、数字写法经常不一致。人工阅读时不太显眼，但做排版发布或者文本分析时会出问题——同一个词在简体和繁体里变成了两个词条，词频统计就散了。中文文本规范化帮你把这些不一致统一起来，处理方式是无损替换，一个字都不会丢。和文本清洗不同，规范化是\"改写\"，清洗是\"删减\"，建议先规范化再清洗。","2026-06-07T07:38:15.167Z","2026-06-07T07:27:11.047Z",{"id":128,"title":129,"type":8,"index":130,"lang":10,"category":11,"tags":131,"coverImages":132,"slug":134,"summary":135,"updatedAt":125,"createdAt":136},"69ecc478135af239496f85ff","文本清洗工具使用教程：批量去除噪声、标点、停用词，输出干净语料",3,[120,13],[133],"https:\u002F\u002Flatte-api-uploader.oss-cn-chengdu.aliyuncs.com\u002Fanna\u002Fadmin-cms\u002F69ecc478135af239496f85ff\u002Fcover\u002F293ee319468feaa66f97d3f7f64820be.png","wen-ben-qing-xi-gong-ju-shi-yong-jiao-cheng-pi-liang-qu-chu-zao-sheng-biao-dian-ting-yong-ci-shu-chu-gan-jing-yu-liao","原始文本里经常混着 HTML 标签、多余标点、emoji、零宽字符这些东西。它们会干扰词频统计、主题建模、情感分析的结果。文本清洗帮你把这些噪声批量去掉，处理完得到干净文本和一份对照预览，方便你确认没有误删。","2026-04-25T13:41:12.691Z",{"id":138,"title":139,"type":8,"index":140,"lang":10,"category":141,"tags":142,"coverImages":143,"slug":144,"summary":145,"updatedAt":125,"createdAt":146},"69eec618d3c481bb3fd2b4aa","TATOOLS 用户隐私政策",4,"privacy",[],[],"tatools-yong-hu-yin-si-zheng-ce","","2026-04-27T02:12:40.234Z"]