Text Intelligence文本智能分析

Tokenise text, extract TF-weighted keywords and compare two documents with cosine similarity.

对中英文文本分词、去停用词、提取 TF 加权关键词，并用余弦相似度比较两段文本的接近程度。

What is Text Intelligence?文本智能分析 是什么？

A text-intelligence workbench built on classic NLP techniques rather than a black-box model. It tokenises mixed Chinese/English input (Chinese by character plus bigrams, English by word), removes stop words for both languages, ranks the remaining terms by term frequency, and — when you paste a second document — measures how similar the two are with cosine similarity over their term-frequency vectors. It is a fast way to sanity-check keyword extraction and similarity before you wire up a heavier pipeline.

一个基于经典 NLP 技术而非黑盒模型的文本分析工作台。它对中英混排文本分词（中文按字切并扩展成二元组，英文按词切），去掉中英文停用词，按词频给剩余词条排序；再粘贴第二段文档时，用词频向量的余弦相似度衡量两段文本有多接近。这是一个在接入更重的流水线之前，快速验证关键词提取与相似度是否靠谱的工具。

How to use text intelligence如何使用 文本智能分析

1. Paste the main text into the first box.
2. Optionally paste a second document into the comparison box.
3. Press Analyse to get token counts, top keywords and the cosine similarity score.
4. Copy the keyword list if you need it downstream.

1. 把主文本粘贴进第一个输入框。
2. 可选：把第二段文档粘贴进对比框。
3. 点"分析"，得到词元数、高频关键词与余弦相似度。
4. 需要时一键复制关键词列表。

Common uses常见用途

• Find the dominant topics of a long article or support thread
• Compare two drafts and see how much vocabulary they actually share
• Sanity-check keyword extraction before feeding a search index
• Pre-process text before sending it to a model, to cut tokens

• 找出长文或客服会话里的主要话题
• 比较两版文稿，看它们真正共享了多少词汇
• 在接入搜索引擎前先验证关键词提取效果
• 送给模型前先做预清洗，减少 token 消耗

FAQ常见问题