Chapter 41: The Shadow Behind It
"Based on the initial speculation, we had assumed that the three cases were interrelated, and if we started from this perspective, I am afraid it will not be difficult to guess who this master is." Chen Xi said.
"A person who is well versed in internet technology... Could it be Feng Ming? He seems to be an executive at a tech company. Following Chen Xi's train of thought, Tang Yuran boldly guessed.
"That's right, that's what we assume, so now we need to gather evidence to test our hypothesis."
"What do you need me to do?" Tang Yuran asked.
"I want to ask, if the police can obtain the chat records of Qin Kai and Feng Ming at the same time, can you tell if the chat partner on one side of them is another person?" Without answering the rhetorical question, Chen Xi threw another question to Tang Yuran.
"Yes." Listening to Chen Xi's question, Tang Yuran nodded firmly. "This kind of thing can be done with enough data."
"Can you tell us about the specific principle of judgment?"
"Absolutely." After Tang Yuran finished speaking, he first paused, and then said: "Actually, the principle is very simple, and it is not difficult to understand. As you just said, as long as you can get the chat records of the two people and collect the text information on them, you can extract the language style and text features of the two people, as long as you have these features, you can construct a model according to the features, to put it simply, this model is similar to a mold. After that, if you want to determine whether a certain text was written by this person, you only need to extract the features of the text content, compare it with the model that has been formed, and get a corresponding matching degree, and then make a judgment based on this matching degree. ”
"So, the better the match, the more likely the text is to come from the same person, right?" After listening to Tang Yuran's explanation, Chen Xi couldn't help but feel a little dizzy, she just listened to a rough idea, but she understood the basic principles.
"That's right, so to speak, this method is probability, and it can't be 100% consistent, so it can only be used as a reference."
"But if the match is high, there will be no problem, right?" Chen Xi asked rhetorically.
"Theoretically speaking, even if someone deliberately wants to imitate other people's style and falsify the content of the text, this kind of thing cannot be created by trying to fake, and there will be many factors involved, which are called characteristic factors. If the imitation is not like it, or it fails to grasp the key point, it will be indecent, and the computer can recognize it at once when extracting features. ”
"It sounds very professional, but I didn't expect the things you studied to be so technical." Chen Xi couldn't help but sigh and said.
"It's not that exaggerated, I'm just a junior in this industry." Listening to Chen Xi's praise, Tang Yuran suddenly felt a little embarrassed, and hurriedly said humbly.
"You don't have to be modest, you have a specialization in the art industry, but I still want to ask, according to your experience, will the accuracy of this kind of computer statistics be very high?" Chen Xi hesitated for a moment, but still expressed the concerns in his heart, after all, this kind of thing is not a little sloppy.
"I understand what your concerns are, but I am very responsible to tell you that the technology I just mentioned using textual features to solve crimes has been in use in many developed countries as early as the nineties. For example, in countries like Japan, their criminal police have long begun to use the text messages left by suspects to solve cases. ”
"It started in the nineties ... So how exactly did they solve the case? Hearing this, Chen Xi couldn't help but sigh, the tools that others had long been familiar with and properly used had come to her, and because of their ignorance, they still questioned it.
"Well, this technology started late in China, and in the academic world, there are not a few people who have studied it, but there are still a few who really put it into practical use, so it is normal for you not to know about it, after all, it takes a certain amount of time to apply it from theory." Tang Yuran said so, as if he was comforting Chen Xi.
"Can you give us some specific examples? I want to learn more. Chen Xi said, as if he had finally seized such a good learning opportunity and refused to let it go.
"Then let me talk about it briefly..."Tang Yuran paused, as if he was thinking, and then said: "In cases like this, text information must be left before they can be studied, so most of the cases related to them are cases disguised as suicides. ”
"How?" Chen Xi couldn't help but ask.
"Some homicides are carefully designed and arranged by the murderer to disguise them as suicides, so that if you want the police to believe that the deceased really committed suicide, you need to leave a suicide note to announce your suicide to the world. Therefore, what the murderer has to do is to forge a suicide note. ”
"I understand, these suicide notes are the object of your research, first extract the text characteristics of the suicide note, build a feature model on this basis, and then extract the characteristics of the text information usually left by the suicide person, compare the two, and get a matching degree, a high degree of matching means that the suicide note is from the hands of the deceased, if the matching degree is low, it means that the suicide note is forged by others. In this way, it is possible to determine whether the deceased died by suicide. ”
"That's right, that's how it works." Tang Yuran nodded in agreement. "In the 90s, in Japan, there was a case in which the police were successfully found to be the murderer with this kind of analysis. In that case, the police suspected that the deceased did not die by suicide, but from homicide, but no matter how they searched, they could not determine who the suspect was, and everyone seemed to have a motive, but because of the lack of evidence, it was difficult for the police to make a judgment for a while, so someone proposed to use this method to find the suspect. ”
"And then?" Chen Xi listened very seriously and hurriedly asked.
"Later, the police collected the text information that everyone usually wrote, studied it, and finally locked up a criminal suspect, and then used this as a breakthrough to track and observe him more closely, and finally they found key evidence, so as to bring him to justice."
"I didn't expect it to be able to do this... If they are all suicide cases, can the three cases encountered this time also use the diary as a breakthrough? As he was talking, Chen Xi seemed to suddenly think of something and spoke.
Listening to Chen Xi's words, Tang Yuran shook his head at first, then sighed, and said: "I have studied those three diaries, and the language style and characteristics of each diary are the same, and there is no such thing as someone else writing them. ”
"And what if someone deliberately imitates it?" Chen Xi couldn't help but ask rhetorically.
"This kind of thing, even if you want to do it, is very difficult, I divided each diary into ten equal parts, selected one of them as the training set, the other as the test set, the results show that each test set and the training set of the degree of matching are similar, if it is really someone deliberately imitated the writing, then the degree of similarity of the match, is inevitably too high." Tang Yuran analyzed.
"I see, so these three diaries should all be from the hands of the deceased, not others, right?"
"That's right." After Tang Yuran finished speaking, he nodded heavily.
"It looks like this idea is not going to work." Chen Xi said, with some regret in his tone.
"Although this method cannot be used in a diary, this conclusion confirms the credibility of the diary." Listening to the conversation between the two all the time, Wei Zhongwen on the side spoke.
"That's right, that's exactly what it looks like."
"Well, that's right."
Hearing Wei Zhongwen's words, Tang Yuran and Chen Xi, both of whom expressed their approval.
"But do you have any data collected now?" Tang Yuran asked.
"Not yet, this is the idea we just came up with, before acting, please come here, just want to ask about the feasibility of this plan." Wei Zhongwen explained.
"This plan, from my personal point of view, is feasible and operational. If it goes well, the results will be available in a day or two, but only if your data is credible. Tang Yuran responded.
"Credibility, what do you mean?"
"Language style is a very abstract concept, and if someone notices your actions and forcibly changes their style, it will lead to a result that has no reference value." Tang Yuran prompted.
"Of course, in the process of our actions, we will try not to startle the snake, which is also our first principle, so there is no need to worry about this." Wei Zhongwen said very firmly as if he wanted to dispel Tang Yuran's doubts.
"Okay, in that case, we can be regarded as having a clear division of labor, I'll just wait for you to give the data, and I will provide you with the data results as soon as possible." Tang Yuran said frankly.
"There's one more thing we'd like to ask you about." Wei Zhongwen looked at Tang Yuran on the side and spoke.
"What's the matter?"
"We investigated the flow of Xu Ziqing's bank card and found a very strange thing."
"Strange things?" Tang Yuran couldn't help but ask rhetorically.
"That's right, didn't you say it before, Xu Ziqing has always loved luxury goods, and often buys some brand-name bags."
"So, what's the problem?"
"According to her bank statement, we found that in her bill, there were indeed several large expenses, but they were not high enough to reach the price of luxury goods. Moreover, most of the payees corresponding to these high-value bills are merchants of some online stores. ”
"What do you mean..."Tang Yuran probably understood, but she didn't say anything, just waited for Wei Zhongwen to continue.