Filtering e-mail messages

[0001] The disclosed system and method relates to the filtering of incoming e-mail messages.

[0002] Spam is the name given to unsolicited e-mail messages that are sent to people indiscriminately. These messages might also be inappropriate. For example, an e-mail message offering cheap contact lenses might be inappropriate if sent to people with unimpaired vision. A contact lens user could, however, consider the message to be of value, even though it was unsolicited. This example illustrates a major issue with such e-mail; only the recipient of a message can truly decide whether it is spam or not.

[0003] Filters are known to check e-mails for spam using several techniques including conventional filters, compliance checking and traffic anomaly detection. For example, filters include software that looks for key words (e.g. XXX, sex, etc.). When a keyword is found the e-mail message is rejected. With compliance checking, a considerable number of tools used by spammers generate non-compliant e-mail messages. These can be identified and ignored. Traffic anomaly detection involves monitoring where e-mail messages come from and go to, which helps to detect unusual patterns of e-mail that are often associated with mass mailings. Software can be configured to take action when specific limits are reached. All of these known filters have several problems including maintenance and false positives. Maintaining a word list for filters can become a full time job for the systems administrator. A false positive occurs when a normal, legitimate message is incorrectly identified as spam. Filtering on a keyword such as breast will catch all discussions on breasts including those concerning say, cancer treatment.

[0004] FIG. 1 is a diagram illustrating e-mail communications between Martin and Joe, a new correspondent according to an example of the disclosed system and method;

[0005] FIG. 2 is a diagram illustrating e-mail communications between Martin and Sam the spammer according to an example of the disclosed system and method;

[0006] FIG. 3 is a diagram illustrating e-mail communications according to an example of the disclosed system and method; and

[0007] FIG. 4 is a diagram illustrating an e-mail system according to an example of the disclosed system and method.

[0008] The disclosed system and method includes an improved filtering technique for incoming e-mails in which, on receiving an e-mail, an e-mail challenge is sent to the sender of the e-mail requesting that they send an e-mail confirmation that they are the sender of the original e-mail, and in which received e-mails are then processed further according to whether or not they are accompanied by a corresponding confirmation e-mail. Optionally, the disclosed system and method allows for a time period to be set in which the validation e-mail must be sent. If the validation is not received, the message and record of the address will be deleted. If a valid confirmation is received then the address can be flagged as a permanently acceptable address.

[0009] This filtering technique is easy to understand and easy to use. Further, this technique saves time by removing the need for users to review spam thereby increasing the time available to handle legitimate e-mail. The number of false positives are also reduced. Users no longer need to review their in-box when deleting spam. This dramatically reduces the likelihood that they delete legitimate e-mail accidentally.

[0010] The disclosed system and method can be used with any mail client. The confirmation process can be used irrespective of mailbox access method. If a mail server is also configured for POP3/IMAP using, say GLMail, then with any POP3 or IMAP4 mail client (e.g. Microsoft Outlook, Microsoft Outlook Express, Qualcom's Eudora, etc.) can be used in addition to the server HTTP client. Further, there is low management overhead. The system administrator does not have to make changes to the system set up for each user. Each user can choose their own individual settings. Accidental leakage of a user's e-mail address is also limited.

[0011] Commonly SPAM is sent from addresses that are unable to receive e-mail, either because of an invalid reply address or by deliberately disabling the account. Such addresses will be caught by the disclosed system and method with no user action required.

[0012] Confirmation is an additional stage in the mail handling process that requires the original sender of a message to take some action when first communicating. This action verifies that the sender exists and intended the message to be sent.

[0013] Once the sender has returned the verification, the original message is delivered as usual. The recipient can elect to have the sender's e-mail address automatically added to their address book, so the next time a message is received from them the confirmation is not activated. This means that the sender has to go through the process only once.

[0014] While messages are waiting confirmation, they are held in a Quarantine folder that may be accessed by the recipient at any time. Old messages are removed automatically from the Quarantine folder after a number of days defined by the user. If a message is removed from Quarantine without a confirmation having been received then the e-mail address can be blocked.

[0015] As an example, consider Martin and a new correspondent, Joe. Martin has just switched on confirmation for the first time. Having previously exchanged e-mail addresses with Joe at a recent conference, Joe now e-mails him for the first time (e-mail 1 FIG. 1). On receipt at Martin's server, Joe's e-mail was diverted into the Quarantine folder. Joe is then immediately sent an automatically generated e-mail from Martin's system (check). The e-mail may look like this:

[0016] Hi

[0017] Many thanks for your e-mail. Please reply to this message.

[0018] Martin

[0019] To explain: I have set up an anti-spam filter which asks for a confirmation from you that you intended to e-mail me. When you reply, the message will be delivered to me as usual and I can read it. You will only have to do this once.

[0020] In order for your message to be delivered to martindance.org.uk, please reply to this message, leaving the following lines intact.

[0021] ---start

[0022] token:3c89cb835b4a941836b94bd6841f60d5:IoZ7J)Sd

[0023] ---end

[0024] On receipt, Joe simply replies to the message (yes, it's me) and does not have to type anything into the body of the message.

[0025] The reply arrives at Martin's account where the confirmation process identifies the token and automatically transfers Joe's e-mail into Martin's In-Box. Martin can now reply to Joe in the usual way (reply, e-mail2, etc.). The token may be in the body, subject or in part of the return address.

[0026] In an alternative example, illustrated in FIG. 3, the e-mail challenge sent by Martin requests that the recipient, in this case Joe, responds by selecting a unique URL (website) to confirm that they are the sender of the originally received e-mail. By selecting the URL the token embedded in the e-mail challenge is automatically passed, together with Joe's e-mail details, to the URL host and from there to Martin's server. This is accomplished by conventional means, such as Java applets.

[0027] In a further example, the e-mail challenge is sent from a unique e-mail address to which a response must be sent to avoid rejection of the original message. Each message that is received by Martin's e-mail server from a previously unknown source causes a new e-mail account to be generated from which the challenge e-mail is sent. Optionally, the unique e-mail addresses may be arranged to expire after a certain period of time. In either case it is not necessary, although it is not precluded either, to include an identifying token in the e-mail challenge as any reply to the unique e-mail address must have been sent from the original sender, i.e. Joe, as only they have been provided with the unique e-mail address.

[0028] There are several choices that Martin has about the way he can use the confirmation process. For example, at any time, Martin can check the special folder called Quarantine. This folder contains all the messages waiting for the confirmation response from the sender. After a chosen number of days, messages are removed from the folder automatically. When messages are removed, the sender's e-mail address can be added to the block list so messages from this person are never accepted again. Martin can also choose to block, accept or delete any messages at any time at his discretion. Additionally, Martin can edit the confirmation message so he can provide a personal response in his own language and style.

[0029] If desired, addresses can be confirmed only once. Confirmation messages are only sent to those people who do not have an entry in Martin's address book. Martin could add e-mail addresses to the address book manually. To ease management, Martin can decide how which addresses are added automatically. For example, every address Martin e-mails can be placed into the address book and/or the address of anyone who goes through the confirmation process. This means that the confirmation does not affect those people that Martin already communicates with. Martin can also decide to clear his address book at any time, thus forcing a reconfirmation in the future.

[0030] Consider now the example of Sam, the spammer. Sam has just put together a mail shot to 10,000,000 people (e-mail 1 in FIG. 2). In the past, Martin's in-box would have received an e-mail containing yet another get rich quick idea. Now, with the confirmation process switched on, he is going to be spared from the hassle.

[0031] When Martin's system receives the message from Sam, it does not find a corresponding address book entry and so sends a confirmation message to the address in the e-mail message (check). One of two things is now likely to happen, depending upon how Sam sent his spam.

[0032] Most spam is sent from accounts that do not exist. In this case, the mail system will return a message (possibly from the postmaster) saying that the account is not available. Since the details of this message do not match Sam's original message, the confirmation process is activated again and both messages reside in the Quarantine folder. Eventually, the messages will be deleted and the spam never reaches Martin's in-box.

[0033] If Sam sent the message from his own account he will need to personally reply to the message in order for it to be delivered to Martin. For a large number of responses, this is clearly not worth the spammer's time.

[0034] The confirmation process can be used with Microsoft Outlook, Eudora, Pegasus, etc. The Confirmation process works on the server by holding messages in a special Quarantine folder and then transferring it to the standard In-Box once a confirmation has been received. If the mail server is equipped with a suitable complement, such as GLMail or NTMail, then any mail client that can use POP3 or IMAP4 can subsequently obtain the message in the usual way. Further, those people using an IMAP4 client can also review messages that appear in the Quarantine folder. Each user need only log on GLWebMail XT once to set up their confirmation options and retains the option to access their mailbox via the inbuilt GLWebMailXT client at any time.

[0035] False positives are only likely to occur if the original sender (in this case Joe) never replies to the confirmation message and even then, only if it is the first ever communication. In this case, Martin is likely to know about the e-mail message through other means (e.g. because they met at the conference) and can simply check the Quarantine folder.

[0036] Experience has shown that confirmation actually reduces the number of false positives. When users receive a large number of spam messages (e.g. five or ten each day), they tend to simply press the delete button to delete anything that looks like spam. This may lead to legitimate messages being deleted carelessly or accidentally.

[0037] For example, Martin might have sent a payment for some new software. The company sends the key via e-mail with a subject title of Amazing Slow Downer for Windows from an addressee of Bernadette. Since Martin doesn't recognise the name, he simply deletes the e-mail. A couple of weeks later, he rings up to find out what happened to his software

[0038] FIG. 4 illustrates schematically an e-mail system according to an example of the disclosed system and method. The new friend or potential spammer 2 is shown in communication with a data processor 4 that is resident in the e-mail system. The data processor 4 is in communication with two data storage areas. The first data storage area 6 stores a list of approved senders, whilst the second data storage area 8 stores a list of blocked senders. The data processor 4 is arranged to access the first and second data storage areas in response to receiving an e-mail from the sender 2 as part of the process, as described hereinbefore, of determining how to process the received e-mail. Cleared e-mails are displayed in a conventional manner in a system in Box 10.

[0039] Although preferred examples have been disclosed for illustrative purposes, those of ordinary skill in the art will appreciate that the scope of this patent is not limited thereto. On the contrary, this patent covers all systems and methods falling within the scope and spirit of the accompanying claims.