Thesis Character Counter
Characters without spaces · mixed Chinese-English breakdown · plagiarism-check fee estimate · nothing is uploaded
📖 How to use it
- Paste the thesis text — chapter by chapter works too — and the counts update as you type.
- Enter your provider's price per 1000 characters to estimate what this check will cost.
- Use the breakdown table to see whether English terms or punctuation are inflating the character count.
- When a report's "total words" disagrees with your own count, work out which column they used before you start cutting text.
❓ Frequently asked questions
Q: Do plagiarism-check services charge by words or characters?
A: Most charge by characters and exclude spaces, while some reports show both a total word count and a total character count. All three measures are listed here; the fee estimate is "characters without spaces ÷ 1000, rounded up × price", with the price supplied by you.
Q: Why does this differ slightly from the report I received?
A: Services apply their own rules to spaces, full-width punctuation, formulas, figure captions, references and acknowledgements, so their numbers follow their rules. The counts here are generic and meant for estimation and self-checking, not as a replacement for the report you paid for.
Q: Is my thesis stored anywhere?
A: No. The pasted text is only used in memory for the calculation: no request, no disk write, no local storage. Refreshing or closing the page clears it. That is also why this tool cannot "check" anything for you — similarity checking needs a comparison corpus, which is a different product entirely.
🧮 Where the difference between measures comes from
"How long is my thesis" has several correct answers, because the denominator is defined differently. Take a text with H Chinese characters, P CJK punctuation marks, W English words, N digit runs and C non-space characters. The two common measures are:
approximate word count W_total = H + P + W + N
characters without spaces C = every non-space character counted individually
The gap comes entirely from English and digits: an eight-letter English word is 1 under the word-count measure and 8 under the character-count measure — a difference of 7. In a mixed Chinese-English thesis the character count therefore runs well ahead of the word count, and by exactly how much depends on the share of English. The row above computes that gap from the text you pasted.
A similarity score is the same kind of fraction: similarity = matched characters / total characters × 100%. Change the denominator and the same matched passage produces a different percentage — one reason two services can report different numbers for one document. What is certain: get the total character count right first, and every estimate after that means something.
🧾 Worked example
Using the sample paragraph:
本文提出一种基于图神经网络（GNN）的推荐算法，在 MovieLens-1M 数据集上将 Recall@10 提升了 4.7%。
| Item | Value | Derivation |
|---|---|---|
| Total characters (with spaces) | 64 | No emoji, so code units = grapheme clusters = 64 |
| Characters without spaces | 59 | 64 − 5 spaces (as in "在 MovieLens-1M 数据集") |
| Characters without punctuation | 51 | 59 − 8 punctuation marks （），。−@.% and friends |
| Approximate word count | 38 | 27 Chinese characters + 4 CJK punctuation + 4 English words (GNN / MovieLens / M / Recall) + 3 digit runs (1, 10, 4.7) |
| Gap between the two measures | 21 | 59 − 38: English and digits are counted differently by each |
| Fee at 1.5 per 1000 characters | 1.50 | ceil(59 / 1000) = 1 unit × 1.5 |
Note the last two rows: 59 characters against 38 words — a gap of 21, which says a lot about how much English and numeric content this short passage carries. Scale that to a 30,000-character thesis at 1.5 per 1000 and a single check costs about 45 — noticeably more than a word-based estimate would suggest, and worth knowing before you submit.
🎯 When this is useful
- Self-checking length and estimating plagiarism-check cost before submitting a dissertation or journal paper.
- Checking a manuscript against a submission system's "no more than N characters" rule.
- Verifying scope before invoicing translation or editing work billed by character.
- Not suited to: claims of matching a specific service's count exactly (their rules are not published), or doing the similarity check itself — no corpus is involved here, and unpublished work should not be uploaded anywhere anyway.