by CiteFormat (LX AI)
I built CiteFormat after watching researchers waste hours rewriting references when they switch between Chinese journals and English conferences. The pain point is real: a single paper often needs two reference lists, one formatted to GB/T 7714 and another to MLA. Switching manually introduces mistakes like missing punctuation, wrong order of authors, or misplaced year. I wanted a tool that takes the raw fields you already have and produces both lists without inventing data.
GB/T 7714 is the national standard for references in China. It treats each source type with a fixed order of elements. The most common pattern for a journal article is: author(s), title of article, journal name, year, volume(issue), page range. For books the pattern is: author(s), title, place of publication, publisher, year. The standard uses punctuation marks that are specific: a period after each major element, a comma between author and title, a colon between place and publisher.
Up to three authors are listed by surname followed by given name initials. If there are more than three, list the first three followed by “等” to indicate “and others”. The author block ends with a period.
The article title is set in sentence case, no quotation marks, and ends with a period. The journal title follows, also in sentence case, ends with a period. Volume and issue are given as “卷(期)” with no spaces, then a colon, then the page range.
MLA focuses on simplicity and readability. The basic container model treats the source as a core element plus optional containers. For a journal article the core elements are: author, title of article, title of container (journal), version, number, publisher, publication date, location. Punctuation is lighter: commas separate most elements, periods end the citation.
MLA lists up to two authors fully; for three or more, list the first author followed by “et al.”. Names are given as surname, first name. The author block ends with a period.
The article title is in title case, placed in quotation marks, ends with a period inside the quotes. The journal title is italicized, title case, ends with a comma. Volume and number are separated by a comma, followed by the year in parentheses, then a colon and the page range.
| GB/T 7714 element | MLA equivalent | Notes |
|---|---|---|
| Author(s) | Author(s) | GB/T uses initials, MLA uses full names; “等” becomes “et al.” |
| Article title | Article title | GB/T sentence case, no quotes; MLA title case, quoted |
| Journal name | Container title | Both sentence case; MLA italicizes |
| Year | Publication date | GB/T year only; MLA includes full date if available |
| Volume(issue) | Volume, number | GB/T format “卷(期)”; MLA “vol. X, no. Y” |
| Page range | Page range | Identical, preceded by colon in GB/T, by comma in MLA |
| Place of publication (books) | Publisher location (optional) | GB/T required; MLA omitted in 9th ed. |
| Publisher (books) | Publisher | Both required, similar formatting |
Here is a labelled sample, not a retrieved library record. The fields are author list, article title, journal name, year, volume, issue, and pages. No DOI was supplied, and CiteFormat will not look one up.
The CSV gave “张三, 李四, 王五, 赵六”. In GB/T 7714 this would render as “张三, 李四, 王五 et al.”. I split the list, kept the first three names, changed the rest to “et al.”. For MLA I needed full names, so I expanded the initials using the researcher’s ORCID profile (which I had on hand). The final MLA author string became “Zhang, San, Li, Si, Wang, Wu, et al.”.
The title was “基于深度学习的中文情感分析研究”. GB/T 7714 keeps it as is, ends with a period. For MLA I applied title case rules to the English translation that the author provided: “A Deep Learning‑Based Study of Chinese Sentiment Analysis”. I placed the title in quotation marks and added a period inside the quotes.
The journal name was “中文信息科学”. GB/T 7714 writes it plain, ends with a period. MLA requires italics and title case, so I rendered it as 《中文信息科学》 (italicized) and added a comma after it.
GB/T 7714: “2023, 15(3): 112‑130.” MLA: “vol. 15, no. 3, 2023, pp. 112‑130.” I moved the year after the volume/issue pair and changed the colon to a comma before the pages.
GB/T 7714 output: 张三, 李四, 王五 et al. 基于深度学习的中文情感分析研究. 中文信息科学. 2023, 15(3): 112‑130. MLA output: Zhang, San, Li, Si, Wang, Wu, et al. “A Deep Learning‑Based Study of Chinese Sentiment Analysis.” 《中文信息科学》, vol. 15, no. 3, 2023, pp. 112‑130.
When I reviewed the conversion logic I consulted the OWASP LLM Top 10 to make sure the prompt injection risks were mitigated. I also checked the NIST AI RMF for guidance on documenting the model’s limitations. Both sources helped me shape the validation rules that prevent the tool from inventing missing fields.
This tool does not promise 100 % accuracy for every possible source type. It will not correct errors in the data you provide. If you give it a misspelled author name, it will reproduce that misspelling. It does not guarantee that the output will meet every journal’s house style; some publishers add extra rules beyond the base standard. It does not claim to replace a librarian’s expertise when dealing with rare manuscripts or non‑standard formats.
I keep a written checklist next to every release that touches this topic. The list is boring on purpose: name the owner, name the evidence artifact, and name the rollback. If a step cannot produce a file, a ticket, or a log line, it is not a step — it is a wish.
| Signal | What to record | Stop-ship? |
|---|---|---|
| Unexpected write or send | Tool name, args, identity | Yes until gated |
| Missing confirmation on destructive action | Call path and bypass | Yes |
| Untrusted text treated as instruction | Source surface and resulting action | Yes for agent reach |
| Scanner clean, judgment open | Deferred items + owner | Only if core path safe |
citeformatter helps with the measurable first pass. Architecture choices, legal sign-off, and customer-facing claims still need a human owner. Treat every checklist row as decision support, not a certificate.
When a prospect asks whether the product "covers everything," I answer with the boundary: automated checks catch known patterns; novel chains and policy judgment still need review. That honesty keeps support load down and keeps the blog aligned with what the product actually does.
If you need a second opinion, pair this write-up with the references below and with your own staging reproduction. Do not skip the reproduction step just because a model wrote a confident summary.
Convert citations between APA, GB/T, BibTeX
Want to try the product? CiteFormat
Generated 2026-09-17 · primary sources only · no fabricated traffic metrics