php - 从 MySQL 数据库表中删除垃圾邮件

Question

我目前正在开发我继承的 PHP 网站。它包括评论功能，当前的评论数据库中有一堆垃圾评论，所有者希望将其删除。

垃圾评论通常包括我当前正在匹配和删除的原始 HTML 或 BBCode URL(评论不允许/呈现 HTML)、带有典型关键字的评论和名称(免费、购买、优惠券、贷款、cialis、伟哥等)和胡说八道，举个例子:

they will cross check with Hotel.After long deliberations she fllainy said that the concerned peson who did booking had leftthe hotel job and now I can be given alternate hotel.I had specially booked this hotel which is right on beach after paying so much money but I was offeredan alternate hotel in the city of Panjim from where to reach beach it takes around than 45-60 minutes and also the fun of staying on beach of my family waslost.Any ways I made the compromise and settled in new hotel in Panjim.But on 28th when I checked in new Hotel Royal PArk Beach Resort I was stunned to see the state of so called 3 star hotel.The hotel was no where near the mentioned status while selling the package.Hotel quality was totally low grade. It better matched a 3rd class sub standard small town inn with dug up entrance,cheaply done up reception area,stinking rooms & bathrooms,burnt curtains,inferior furniture,ongoing contruction next to my room,dilapidated bathrooms fixtures,dirty floor,dirty and humid bed linens and overall bad feeling.[I have proof of all photographs]I immediately informed make my trip' of state of the hotel and wanted alternate hotel but make my trip' kept on telling me to wait and they will look into it.Since my chlid was sick and immediately needed rest I had no option but to check in at this Hotel.Simultaneously I kept on calling Makemytrip for hours but tono avail.This wasted an entire day.Finally I arranged a new genuine 4 Star Hotel Room Rent for which was Rs. 5500/- per night( much below Rs. 7211/- for thishotel booked by Make my trip).I informed MakeMy Trip on 28th Jan 2012 to cancel my reservation and refund me entire money.Also sent a mail to them in this regard.I am still wating for positive response from them but I am least hopeful of getting it.So now I am approaching court to get my dues and compensation.Arvinder

我一直在进行字符串匹配和删除明显的术语，但这有点乏味。如果有人有有效的方法来做到这一点，我真的很想听听。

Answer 1

我有一个与我关心的应用程序类似的问题。我将提供垃圾数据的人称为“ clown ”，并且正在收集一套识别潜在 clown 的规则。最初的规则集非常基础(可能类似于您现在拥有的关键字)。根据我自己的观察和同事的提示，这套规则随着时间的推移而扩展。

请注意，我声明我识别潜在的 clown 。可能包含 clown 数据的表有一个 IsClown 列，它可以表示“未检查”、“潜在 clown ”、“不是 clown ”、“是 clown ”等状态。我有一个检查所有新数据的自动过程。如果该过程匹配一个或多个规则，该条目将被标记为“潜在的 clown ”，并且人类使用简单的 UI 检查条目，将他的状态移动到“不是 clown ”或“是 clown ”。

这是否适合您取决于规则的定义程度以及您正在考虑的数据量。它对我有用，所以我想我会分享。