維基百科:優良條目評審指南

優良條目是指行文、事實準度與出處、覆蓋面、穩定性、多媒體材料使用等方面達到一定品質底線（即優良條目標準）的條目。就如何依標準公正評審優良條目的問題，下文將提出建議。

關於流程

英文維基百科的優良條目（GA）評審流程旨在簡潔。評審不設多人投票、無需尋求共識，也不設評審委員會；所有人均可提名條目，且無關的任何註冊用戶都可以接手評審。

流程仰賴評審人的正直。評審人不得接手自己參與編輯的條目。評審應專注檢查條目符合優良條目標準與否，不應為各種因素所左右，例如：條目怎樣才能更「完美」、評審人自己會怎樣寫、評審人自己對主題的想法如何^[1]。評審人應專注就內容和形式提出建議，而不應強推自己的偏好。

評審人要批判性閱讀條目，根據優良條目標準公正地評判。如果評審人認為條目滿足標準，則應支持其入選優良條目。如果條目未滿足全部標準，評審人有兩種做法：

同優良標準有些許差距——通常情況下，請在評審區寫份評論，詳述條目不符合哪些標準，怎樣改才能達標。如果問題微小或能輕鬆修正，歡迎評審人自行動手。
問題重大或泛濫——可按即時不合標準結案。但即時不合標準的原意是應對草率提名（因為提名門檻很低），因此除非面對「路人式」提名（即提名者無意回應評論），否則不應動用此舉。即時不合標準或會激怒熱衷改善條目的編者。優良條目標準中詳細說明了即時不合標準的要求。

提名者是在優良條目提名頁上提交參選條目的人。提名者應懷抱改善條目的目標，確認條目遵循格式手冊和相關專題的指引，並讓條目滿足優良條目標準的所有要求。關提名者的詳細介紹，可參見優良條目提名指南一文。

Why articles are nominated for Good article status

Nominations are occasionally made by editors who happen upon an article that they believe is good quality, but the majority are made by editors who have spent extensive time working on the nominated article and are interested in improving it.

The Good article process is one of Wikipedia's systems for providing a detailed review of an article. Regardless of whether the article meets the criteria, most of the editors of a nominated article appreciate receiving concrete, specific suggestions about how the article could be improved. If making suggestions about how to improve the article, reviewers must clearly differentiate between improvements necessary to meet the actual standard Good article criteria and ideas that might help the article meet A-class criteria or featured article criteria.

How to review a Good article nomination

The process and mechanics of reviewing a nomination are described by the Good article nomination instructions page, Reviewing section. The guidelines below focus instead on the evaluation and decision-making involved in reviewing a nomination.

First things to look for

Before conducting an extensive review, and after ensuring you are viewing an unvandalized version, check that the article does not have cleanup banners that are obviously still valid, including {{cleanup}}, {{POV}}, {{unreferenced}} or large numbers of {{citation needed}}, {{clarify}}, or similar inline tags.^[2] If it does, you are entitled to fail the article without reviewing further. You can also give the nominator a chance to address the tagged issues if you wish. You should also check to see if the article has been nominated previously and if there are any outstanding issues from the last review.

If while reading the article you think that it falls a long way short of the six criteria you have the option of making a note of how it doesn't meet the criteria on the review page and failing the article without placing it on hold. The reviewer has some discretion when it comes to this option, but it should only really be used when they think the issue can not be solved in a reasonable amount of time or without major rewrites.^[3] If at any time you find significant close paraphrasing or copyright violations the article can also be failed without further review.

Assessing the article and providing a review

If a nomination requires a detailed review, there are two basic steps: assessing the article against the Good article criteria, and communicating this assessment to other editors. For both steps, you may find it helpful to use a checklist, such as {{GAList}}, {{GAList2}} or {{subst:GATable}}; however, this is not required, and if you prefer to communicate your assessment purely in prose, do so.

Step 1. As you read through the article, keep in mind the Good article criteria.

The article should be clearly written, in good prose, with correct spelling and grammar. Check for coherent formatting, good organization of the article into sections, appropriate use of wikilinks, and other aspects of the Manual of Style referred to in the Good article criteria. After you have read the article, check that the lead section is a good summary and introduction to the topic.
The article should be factually accurate according to reliable sources, with inline citations (typically using footnotes) for the six types of material named in the GA criteria.^[4] The article should not copy text from sources without quotation or in text attribution, and it should not contain any original synthesis of source material, or other forms of original research. Perfectly formatted citations are not required. Read the detailed guidance at Wikipedia:DEADREF before addressing any non-functional URLs.
The article should broadly cover the topic without unnecessary digressions. The article may, and sometimes should, go into detail, but it is not required to be comprehensive.
The article should be written from the neutral point of view: this viewpoint strives to represent all other views fairly, proportionately, and without bias. Ensure that the article describes disputes without engaging in them.
The article should be stable, with no ongoing edit wars: constructive article improvement and routine editing does not apply here.
The article should comply with image use policy. Images are encouraged but not required. Any images used should be appropriate to the article, have captions and free licenses or valid fair use rationales.
The article is free of obvious copyright violations. Reviewers can use several tools, as well as Google searches, to help establish whether material has been plagiarised or cut-and-paste from some of the electronic sources used; but this is not a trivial undertaking.

Ideally, a reviewer will have access to all of the source material, and sufficient expertise to verify that the article reflects the content of the sources; this ideal is not often attained.

At a bare minimum, check that the sources used are reliable (for example, blogs are not usually reliable sources) and that those you can access support the content of the article (for example, inline citations lead to sources which agree with what the article says) and are not plagiarized (for example, close paraphrasing of source material should only be used where appropriate, with in text attribution if necessary). If you can not access most of the references you should confirm the most important content of the article via alternative means. Reviewers can confirm information from sources they cannot access at the resource exchange or request translations at Wikipedia:Translation. Beware of on-line sources that have copied content from Wikipedia.
Copyvios and plagiarism might be found by detailed checking of referenced material against the declared sources used in the article, but is more difficult in those cases where undeclared sources have also been employed. For book sources: digital images of some out of copyright books may be found through ebook sites, electronic archives, such as Project Gutenberg, and on Google books. For books still in print, Amazon in its many forms (i.e. by country, such as: .com, .ca, .co.uk, .fr, etc.) has selected images of some copyrighted books, where the copyright holder has given permission, and this may vary from country to country; otherwise the books themselves may need to be consulted in a library and/or borrowed. Some journals and newspapers have also been archived and are available in electronic form, such as The Times, The Washington Post, etc. The use of some of these sites may be free, provided that you register with them; others will be subscription only, but some public library borrowing cards and/or university library cards do provide free access for their members to certain sites of their choosing. Google internet searches can also be used by copying sentences from the article and putting them into the search box. If any are found, the Duplication Detector can be used to detect instances of plagiarism or copyvio. Here the reviewer needs to enter the article's url and the source url into the two fields and the software compares them. It is usually best to start with sentences that appear "out-of-place", though a few sentences should be checked at a minimum. Remember that a match may indicate that the other website copied Wikipedia, rather than the other way around.

Step 2. Summarize your assessment on the review page, noting any problems you have found, preferably with specific examples. To avoid conflict and encourage other editors, try to write your review in a positive way, raising problems as opportunities for improvement rather than criticism of previous efforts; suggesting specific fixes can be particularly helpful to other editors.

If the article meets the criteria, it should be listed as a Good article; bear in mind that future editors may be interested in your reasoning, so don't just leave an all-positive checklist. Suggestions for further improvement may be welcomed by article editors.
If instead, the article has problems, you need to decide whether to fail the article (leaving a detailed review), or place it on hold. Many of the considerations concerning "quick-failing" apply here. As the nominal hold period is one week (this is a suggestion, not a requirement) one rule of thumb is to consider whether the issues you have raised could reasonably be fixed within such a timescale. If that is possible, then the article should be placed on hold. If you decide not to place the article on hold, ensure your review provides detailed advice for improvement, and encourage editors to renominate the article once the problems have been fixed.

Step 3 Indicate the result by marking the review as passed or failed, placed on hold, or requiring a second opinion, as appropriate.

Mistakes to avoid in reviews

Giving problems, not solutions

When reviewing an article, do not just describe its shortcomings, provide suggestions to fix them. For example:

"The lead is messed up" does not give editors any guidance to make the lead better. Instead, try something like. "The lead does not adequately summarize the article. Try expanding it. See Wikipedia:LEAD for more information".

Using the Good article criteria as a guide to organize a review is a good thing, but do not merely use it as a checklist. For example:

"This article violates criterion 1 of the Good article criteria" with no further information does not help anyone improve the article. Instead, try something like: "This article is dominated by its plot summary and takes an "in-universe" perspective. According to the Manual of Style, an article about a work of fiction should be written primarily from an "out-of-universe" perspective. That needs to be fixed before this can be a Good article."

Imposing your personal criteria

When reviewing, focus on providing the best review you can for that article. Take care not to be distracted by whether related top-level articles are lackluster, or let frustration over how many articles there are covering a given topic area show through.

Avoid commenting on the perceived "merit" of the subject of the article. If an article on a porn star is well-written, well-organized, well-referenced, and follows the specific content policies and style guidelines that are required by the GA criteria, then you should not fail it because you think Wikipedia has too many articles on porn stars.

Enthusiasm in wanting an article to be the best it can be is admirable, but take care not to impose conditions for passing the article, perhaps based on your own stylistic preferences, that exceed the criteria. In particular, the GA criteria do not require compliance with several major guidelines, including Wikipedia:Notability and the main Wikipedia:Manual of Style page. As you are determining whether an article is complete, well-written, properly sourced, etc., you may find it useful to consult various community-wide guidelines or advice pages from WikiProjects. These guidelines are often very helpful, but be careful that you do not wrongly require compliance with any guideline that is not specifically mentioned by the Good article criteria. For example, reviewers must not fail an article over the presence or absence of an infobox, even if a relevant WikiProject has declared that infoboxes are extremely important (as some have) or extremely undesirable (as others have) in articles within their scope.

Passing articles that do not meet the Good article criteria

Beware of the several temptations to pass an article that does not meet Good article criteria. For example:

You are a fan of the subject, and want to see the article listed. (Instead, consider improving the article, rather than promoting it.)
You want to illustrate a point.
As Good articles are not Featured articles, it is all right to let things slide. (It is not. The standards for Good articles are less exacting than those for Featured articles, but they nonetheless mean something and should be upheld.)
It appears that the article is as good as it will ever get, and will never meet the standards. (Not every article can be a Good article. If the references to improve an article to Good article standards simply do not exist, then you should not overlook that part of the criteria.)
The editors of the article have obviously spent a considerable effort improving the article during the Good article review process, and even though it does not meet all of the criteria, it is much better than it was when it was first nominated.

Dealing with disputes

While we hope that a Good article review will be a positive, friendly, collaborative experience, disputes sometimes arise. Multiple mechanisms are in place for resolving disputes.

Get a second opinion

If a dispute arises, anyone involved in the review (reviewer, nominator, other editors at the article) can ask for a second opinion.

Leave a clearly labeled note on the review page that specifies what you'd like from the person giving a second opinion. Do you have a single, specific question? Do you want an overall opinion about whether the article meets the GA criteria? Try to be as direct and specific as possible.
On the talk page of the article set the GAN status parameter to "2ndopinion" as in {{GA nominee|...|status=2ndopinion}}. The bot will then add a note to Wikipedia:GAN that you're looking for a second opinion.
Alternatively, you can personally request help from an individual, GA mentors, or relevant WikiProject, or use the normal third-opinion process.

Change reviewers

Occasionally, it is necessary to change reviewers in the middle of a review, rather than closing one review and re-nominating the article. Instructions on doing this are given at Wikipedia talk:Good article nominations/FAQ.

Close and re-nominate

If the reviewer is dissatisfied with the article, but the editors of the article are confident that it meets the GA criteria, it is appropriate for the reviewer to close the GA review (as 'not listed'). The editors may re-nominate the article at any time, including immediately. It is generally best for the first reviewer to allow another person to conduct the subsequent reviews. Subsequent reviewers often carefully consider the comments in recent reviews, and the nomination is more likely to be successful if all relevant issues are addressed appropriately before re-nominating the article.

Send it to GAR

Wikipedia:GAR—the Good article reassessment process—can be used for both listing or delisting articles. It is open to articles that have been reviewed at any point in time, but is not normally used for articles whose reviews are currently open. If an article has received a poor-quality review it is usually faster to simply renominate it at Wikipedia:GAN.

Delisting older articles

The Good article process has evolved considerably since it began.^[5] Despite these changes, and indeed as part of its nature, the Good article process does not always get it right the first time: this is the price paid for the efficiency of the one-nominator:one-reviewer approach. Consequently, if you come across an article which no longer meets the criteria, you may remove it from the Good articles list by following the Good article reassessment guidelines.

Useful tools

Checklinks,^{[失效連結]} check external links and citation parameters; read Wikipedia:DEADREF on dealing with any dead links
Contributors, sortable page version history
WPPageHistStat, edit history overview
Duplication Detector,^{[失效連結]} used to check for Wikipedia:COPYVIO and Wikipedia:PLAGIARISM

Notes

^ 相關專題的活躍編者應該特別注意這點。有時條目適合由專家審閱，但其他情況下，最好不要讓評審者與主題距離太近。
^ The tags must be present before you start your review. It is also worth checking the date they were added (just hover over the tag) as the nominator may not have had time to address any recent ones yet.
^ Although there is no set "reasonable amount of time", a week appears to be the generally accepted standard.
^ Small articles that have a single main source may still be adequately referenced without the use of inline citations. Inline citations may not be required for some articles; the criteria name the only six types of material that require inline citations.
^ Good articles began on 11 October 2005, and the nominations system was introduced on 10 March 2006. (See Good article statistics.) During 2006–2007 the Good article criteria were refined and improved, and during 2007–2008, processes were changed and review pages introduced. A major sweeps effort has ensured that old Good articles meet the current criteria, which have essentially been stable since 2008.

[1] 相關專題的活躍編者應該特別注意這點。有時條目適合由專家審閱，但其他情況下，最好不要讓評審者與主題距離太近。

[2] The tags must be present before you start your review. It is also worth checking the date they were added (just hover over the tag) as the nominator may not have had time to address any recent ones yet.

[3] Although there is no set "reasonable amount of time", a week appears to be the generally accepted standard.

[refs-4] Small articles that have a single main source may still be adequately referenced without the use of inline citations. Inline citations may not be required for some articles; the criteria name the only six types of material that require inline citations.

[5] Good articles began on 11 October 2005, and the nominations system was introduced on 10 March 2006. (See Good article statistics.) During 2006–2007 the Good article criteria were refined and improved, and during 2007–2008, processes were changed and review pages introduced. A major sweeps effort has ensured that old Good articles meet the current criteria, which have essentially been stable since 2008.

[1]

[2]

[3]

[4]

[5]