Обсуждения » Обсуждение сайта Greasy Fork
New script search
Chinese university online learning platforms, such as www.chaoxing.com, www.zhihuishu.com, and www.icve.com.cn, have keywords like "超星" "学习通" and "智慧树",These platforms closely monitor student use of third-party software and scripts, frequently updating their site functions. Open-source scripts are particularly scrutinized and often detected by the platforms within weeks of their release.
When using the new search function (e.g., https://greasyfork.org/zh-CN/scripts?new=1&q=%E8%B6%85%E6%98%9F) to look up these keywords, the top-ranked scripts are usually those updated months or even years ago. Some scripts with limited functionality that do not significantly impact the platforms might still be usable. However, these scripts generally lack maintenance, and the authors may have ceased updates, rendering most scripts ineffective and potentially harmful to users.
The results you're seeing are because the order is based on solely the position of the text in the script. Scripts that have the exact text at the start of their names come up first, and I believe that shorter names that still contain the text end up higher.
It's possible to adjust the order based on other factors. For example, if I boost based on the number of daily installs, the top three results for "超星" would be:
- 🥇超星学习通｜知到智慧树--网课小助手|修复视频播放|自动跳转任务点|自动答题|超高题库覆盖率|逐渐支持更多平台
- 🔍一步搜题，自助查题｜无侵入，更安全｜超星学习通｜智慧树｜成人本科｜继续教育｜复制、截图搜题｜超全题库，专业开发，值得信赖
- 💯【超星学习通满分助手】支持任务点自动跳转|章节测验、作业、考试全网检索答案，简答题支持chatgpt对接|音频、视频全自动静音播放|可视化参数配置
While the current #1 (超星允许粘贴) drops all the way to #61.
This boost could also be done based on reviews, last update date, etc. and be made stronger or weaker. I do want to avoid having the effect that the top results stay at the top just because they're the top result, but I understand what you're saying about outdated scripts. Ideally these would be removed by having "bad" reviews posted on them but that doesn't always happen.
I've made an adjustment so that it boosts results based on the reviews received. This metric isn't perfect but I think it will give better results than before and without leading to the install count loop that would happen by boosting by installs.
Top 3 are now:
which all have the search term in their name and a small number of mostly positive reviews. If these scripts are indeed non-functional, what I expect (hope) to happen is that they'll get extra installs for a while, then some bad reviews, then drop in the rankings, and possibly get removed altogether if they trigger the consecutive bad review process.
I can also make this boost stronger or weaker depending on how we think those results are.
Thank you very much for adjusting the script ranking mechanism. It should now be able to avoid the negative impact of the "Matthew Effect" on installation volume. Most users who search these keywords like to install user scripts in bulk. If a script doesn't work or is too complicated, they will immediately install the next one, resulting in a large number of installations, this is unhealthy. As far as I know, there are many similar scripts using these keywords. Some are just simple modifications of another script with added copyright information, released and then no longer maintained.
The demand for these online learning website scripts has always been high. If users visit these websites without using scripts, it becomes "unbearable." Learning the courses on these websites using normal methods takes tens of hours, and users cannot do other things during learning because these websites restrict mouse and keyboard usage. Learning is good, but the quality of the resources on these websites may be outdated, and there is not much to gain from genuine learning. Moreover, schools do not emphasize online course learning enough, even teachers don't truly understand these courses. These online courses take up a lot of users' time without benefiting their studies. Therefore, users recommend these scripts to each other, which explains the high volume of scripts and users.
Typically, these users do not leave reviews unless they need new features added or if previously working scripts become ineffective. If a script only provides a simple page optimization, there may be no feedback when it stops working. However, if a script negatively impacts them, such as causing low grades or being flagged for cheating, users will definitely leave negative feedbacks.
In the end, thank you for understanding my feedback and making modifications to the website. I believe the user experience on GreasyFork will improve.
Jason, would you add the field like category or tag to let the author describe their userscript better?
For example, scripts for videos sites belongs to "entertainment", scripts for io games belongs to "gaming", scripts like AutoPaging and Allow right clicks belongs to "productivity" or "tool".
If you allow the authors choose the category, the search would be easier. Filter can be also applied as well.
Not really related to the changes being made, but for the use case you're suggesting, it's better for people to just browse by site. I would be open to the idea if there were categories that scripts could easily be slotted into that aren't just proxies for what site they affect.
I've set the new search live, will monitor for stability issues, bug reports, and complaints.
Can we also have the search term "look into" the script description on the script page? (I'm not talking about the // @description)
There are several tools and APIs that can help you categorize websites. Here are a few options:
Klazify: This API uses machine learning to scan a website's content and meta tags, then classifies the site into categories. It supports a wide range of categories and can also extract logos and social media links.
RapidAPI's Website Categorization: This API uses natural language processing to analyze website content and meta tags, assigning categories based on the extracted text.
WhoisXML API: This tool also uses machine learning and natural language processing to categorize websites. It's known for its ease of use and accuracy.
These APIs should help you categorize a list of websites into categories.
Can we also have the search term "look into" the script description on the script page?
It does.
Although I really like this idea, in theory, I 'vote' against it, as authors are just as human as the rest of us, and too many will be tempted to purposely mis-categorize their works.
The closest we have to a 'tagging' system is the Script Sets, which all users can self-create and self-categorize.
Perhaps capture Script Set names thru-out the site (am I correct- each Set gets a unique number?), and publish to a page where everyone can see what others (anonymously?) have tagged; merge Sets with the same name.
NOT requesting this, just throwing out the idea. Can be separated to a separate discussion.
Jason, would you add the field like category or tag to let the author describe their userscript better?
For example, scripts for videos sites belongs to "entertainment", scripts for io games belongs to "gaming", scripts like AutoPaging and Allow right clicks belongs to "productivity" or "tool".
If you allow the authors choose the category, the search would be easier. Filter can be also applied as well.
Add the parameter
new=1to search results to try out an updated script search. This is using a different search engine behind the scenes. Goals are:Please report any problems you see and include the URL you are on.