Triggering keywords?

Another email arrival notification!

Lin Yuan instinctively frowned, 'Did it send repeatedly?'

This kind of situation does happen; after all, any software system is bound to have bugs, especially a software system that covers network communication, where retransmission is common. The biggest trouble in network communication is "I said something, but I don't know if you heard it."

Lin Yuan unhurriedly opened his inbox. However, even before he clicked on the email, just by looking at the subject, he noticed something unusual.

[Automatic Reply: Thank you very much for your letter.]

Although this was still an automatic reply email, the previous automatic reply email did not contain phrases like "thank you very much."

Lin Yuan clicked to view the email.

[Your letter has been received. Our staff will contact you as soon as they are on duty. Please pay attention to email replies.]

The wording of this email was clearly much softer than the previous one, and it no longer emphasized word usage precautions to the sender.

Lin Yuan naturally wouldn't think he was just incredibly lucky and thus received special treatment. From an IT professional's perspective, it was highly likely that the email he had just sent triggered a keyword strategy in the recipient's email processing system.

Using software or AI for email pre-processing is already a very mature technology, as AI intelligent customer service has been widely integrated into the customer service systems of major online shopping platforms.

Lin Yuan then naturally became curious about what keyword triggered this different email reply.

He placed the two emails he sent, one after another, side by side for careful comparison. In the second email, he had only made a tiny adjustment to the line spacing because the other party had emphasized wording requirements, and he had added the signature of Zijin Aeronautics and Astronautics University.

Lin Yuan didn't think this tiny format adjustment would trigger a keyword, so there was only one answer—the signature of Zijin Aeronautics and Astronautics University.

“Heh~~”

Lin Yuan's lips moved slightly; it seemed this organization named OCR indeed had a problem.

Lin Yuan immediately opened his browser and typed into the search bar: "Universities in my country on the sanctions list."

Because he vaguely remembered that the university he was attending should be on the sanctions list.

The search results quickly appeared. Lin Yuan remembered correctly; Zijin Aviation School was prominently listed among the 15 sanctioned universities in Dragon Country.

These 15 universities were sanctioned simply because they were deeply involved in the national defense industry. And using a university deeply involved in the national defense industry as a signature actually triggered an additional email reply from the other party so quickly...

This was too eager. It made Lin Yuan feel: How could the fishing line just start to sway, and the fisherman was already eager to pull the rod?

'After all, there's too little experience.' Lin Yuan became interested. He immediately went online to get a new email account and then sent another email to the other party. (PS: Why would Lin Yuan use an email account linked to his real-name phone number for this kind of thing?)

The content of the email was fabricated, but the signature remained: Zijin Aeronautics and Astronautics University.

Of course, Lin Yuan wasn't so foolish as to send it without changing his IP. He found a free proxy software online that could connect to an HK server, and only then did he send the email.

Although the actual public IP changes when surfing the internet within a local area network, the change has a time cycle. If Lin Yuan foolishly sent emails with his secondary account without using a proxy, the other party might summarize the email source IPs. Once summarized, they would find that emails sent from Lin Yuan's two different accounts either came from the same IP or from IPs with very similar geographical locations.

Then wouldn't that be game over?

Lin Yuan's purpose in sending emails with his secondary account was both to test the waters and out of a mischievous impulse.

Before long, the secondary email account also received an automatic reply email from the other side. But this time, the automatic reply email was not in the style of the extra email; it was the initial, very ordinary email.

Lin Yuan understood after a brief thought.

The other party didn't seem foolish enough to directly search for keywords.

If it were a direct keyword search within the email content, then as long as "Zijin Aeronautics and Astronautics University" was in the other party's keyword list, it would trigger an additional automatic email reply. This is the lowest form of email content evaluation.

But after testing with the secondary account, this was not the case.

Lin Yuan probably guessed that some machine learning algorithms were used, for example: comprehensive extraction of multiple types of keywords, and then using a K-nearest neighbor algorithm or something similar. If it was even more sophisticated, the entire email, from title to content, would be thrown directly to AI for judgment.

Therefore, Lin Yuan concluded: The email he sent, precisely because he added his university's signature at the end, triggered the other party's algorithm to pay close attention.

The content of his email was already very close to the other party's algorithm's G-spot, and the final signature was the last push that tilted the scales directly.

To be on the safe side, Lin Yuan decided to do another test.

He rewrote an email requesting computing resources, but made sure the writing style was different from his previous one, just in case the other party's email retrieval system had a duplicate checking function.

The signature Lin Yuan added to this newly written email was: Qingbei University.

This is Qingbei University, that's prestigious enough, right?

Soon, he sent the email. And just as quickly, an automatic reply was received.

However, it did not trigger an additional email reply. In other words, a computing resource request email with similar content, paired with the signature of Qingbei University, did not excite the other party.

This was getting interesting.

The prestigious Qingbei University actually left the other party indifferent...

Lin Yuan really wanted to try other universities on the sanctions list, but this type of computing resource application system would not receive frequent applications in a short period of time. Because clearly, not many people would know about the OCR organization.

Therefore, even though Lin Yuan could now change his IP using a proxy, he dared not send emails too frequently. Otherwise, receiving several emails from different IPs within the same time frame would inherently be very suspicious.

So, what to do next? Get a good night's sleep.

Presumably, the other party wouldn't write such an algorithm for no reason, as this algorithm clearly gives weight bonuses to universities on the sanctions list.

'The most cunning fishermen often appear in the form of fish.'

Lin Yuan had now gently touched the other side's hook, and the other side had already made a move to pull the rod. The best thing to do next was to wait; at this moment, he absolutely must not be anxious. He absolutely must not put on a posture of "I'll bite even a straight hook."

Lin Yuan did not want to delve into the details of the COR organization. He currently had neither the ability nor the energy for that; he just wanted to leech some computing resources for free. Of course, this process, through "North America jungle"-style tricky operations, would also weaken the enemy. But this was clearly not Lin Yuan's primary goal at the moment.

Lin Yuan even went to look in the mirror before going to bed.

'Am I too righteous? Should I read some "Reader" or "Zhiyin" to refine my personal temperament, so I can gain the enemy's trust in subsequent communications?'

Alright, the protagonist's idleness is over.

As expected, the data once again plummeted. This is truly amazing; two times writing about girls, two times the data was cut in half.

No more big events later. Thanks to the readers who have followed this far...