EP 3 023 948 A1 relates to a method for detecting of a risk of substitution of an authentic electronic payment terminal by a fraudulent electronic payment terminal.
Brief outline of the case
The application was refused and the applicant appealed the refusal.
The board held that claim 1 of the MR lacked IS. The AR were not admitted in the procedure pursuant Art 13(1+2) RPBA. The refusal became final.
In order to detect fraud, it is proposed to use a threshold such that an alert is triggered as soon as the detection parameter, i.e. a number of transactions during a predetermined time period, a cumulative amount of transactions or a combination of both, is below a predetermined threshold. This implies that a fraudulent electronic payment terminal has been substituted to the authentic electronic payment terminal.
The applicant’s point of view
The applicant agreed with the board on the CPA.
The problem formulated by the applicant is “[improving] the reliability of detecting the substitution of a genuine terminal by a fraudulent terminal”.
The applicant considered that the skilled person would not take into account either document D6 or document D4, since these two documents do not disclose comparable fraud detection of a payment terminal such as that known from document D3, or as used in the invention.
In D6, the exact purpose of fraud detection remains undefined, while in D4, it is fraud detection in connection with the authorization of payment transactions and not authentication and monitoring of an individual payment terminal.
The board’s decision
Claim 1 of the MR comprised a step for generating an alert when at least one operating characteristic associated with the authentic electronic payment terminal does not correspond to the authorized values with respect to a predetermined threshold.
As far as the threshold is concerned, the CPA showed in particular how to monitor the relevant parameters using an alarm trigger threshold.
For the board, the OTP formulated by the applicant was not convincing, since the process defined in claim 1 does not appear to be any more reliable than the known process in the CPA.
In both the invention and the CPA, the parameters used may indicate potential fraud, without it being possible to confirm this with certainty.
In both systems, this finding of fraud is based solely on assumptions concerning the values obtained from the parameters used for the indication of fraud in relation to the predetermined threshold, i.e. in relation to the conditions imposed.
For this reason, the board concluded, contrary to the applicant’s arguments, that the OTP solved by the distinctive features did not go beyond the provision of a method with an alternative parameter for detecting potential fraud.
A skilled person seeking further parameters for detecting potential fraud will consult either document D6 or document D4.
These two documents also concern the detection of potential fraud in automatic payment systems in general.
When examining IS, it is not necessary for the systems disclosed in the CPA and the document considered by the skilled person to function identically. If this were the case, it would be necessary to analyse the novelty of the second document in relation to the invention, rather than its inventive step.
What’s more, the skilled person, as part of his general knowledge, a basic understanding which enables him to disregard certain facts. If this were not the case, any reasoning relating to the inventive step would become virtually impossible.
Consequently, when considering the teachings of the CPA, the skilled person would undoubtedly consult either document D6 or document D4, to consider the possibility of including the number of transactions in order to detect potential fraud. The skilled person would conclude that this parameter, the number of transactions, is relevant for detecting potential fraud in the system/process known from the CPA.
The skilled person therefore recognizes, in accordance with the teachings of the CPA, that in the event of disconnection or replacement of the authentic terminal due to fraud, transactions can no longer be counted and the number of transactions decreases, or even tends towards zero.
Consequently, when the skilled person uses the number of transactions to detect potential fraud, he or she uses his or her general knowledge to adapt the threshold so that it triggers an alert when the number of transactions falls below a predetermined threshold.
The choice of the threshold that triggers an alert when the number of transactions falls below this threshold is therefore a direct result of the general knowledge of the person skilled in the art when implementing the parameter proposed in document D6, namely the number of transactions, in the process and/or system known from the CPA.
Comments
The decision is interesting is that it states that when the subject-matter of a claim does not give better results than the CPA, the OTP is merely to find an alternative.
The board made also clear that when examining IS, it is not necessary for the systems disclosed in the CPA and the further document considered by the skilled person to function identically. This would actually imply at least a link between those documents of the prior art.
Furthermore, the skilled person, has as part of his general knowledge, a basic understanding which enables him to disregard certain facts. If this were not the case, any reasoning relating to IS would become virtually impossible. In other words, the skilled peron does not have blinkers.
When starting from the CPA, suggestions going in the direction of the claimed subject-matter can however not be ignored. IS can be denied when the skilled person, on the basis of its common general knowledge, finds suggestions in the available prior art and combines those with the teaching of the CPA.
That the link between the CPA and other pieces of prior art goes via the common general of the skilled person, should however not come as a surprise. Otherwise it would mean that IS should always be acknowledged when there is no direct link between the CPA and the other available prior art.
Comments
4 replies on “T 116/21 – Inventive step and the role of common general knowledge”
Interesting points there, Daniel. Your report brings to my mind the thoughts of the US Court of Appeal on the ways the notional skilled person is deemed to process the common general knowledge. The court declared that the notional person is “not an automaton”. Rather, the person is the possessor of ordinary creativity.
But perhaps with the rise of AI, we shall soon have to retract from saying that the skilled person is “not” an automaton?
Dear Max Drei,
Thanks for your comments. I do not think that the way the US Court of Appeal envisages the way the skilled person is using CGK is different from that of the EPO.
You do however raise an interesting question with respect to AI.
I have just been at a small conference on the effects of AI in IP. One thing seems clear: an inventor will never be a machine, but a natural person.
The problem with AI, under its different forms, neuronal networks, deep machine learning, large language models, it will remain a black box. The result of the black box depends on the training given.
In the absence of any information on the training data and the correlation algorithm we do not have the faintest idea how the result is achieved. This is rather disturbing, not only in IP.
We will at least have to see what AI was able to create as CGK at the effective date of the claim. This might not be as easy as with a wayback machine when it comes to Internet citations as prior art. I have not heard about a wayback machine for AI. This might exclude the use of AI to define CGK after the effective date of a claim.
Only what is, in all certainty, effectively published by an AI prior to the effective date of claim can be taken in consideration as it is prior art, not necessarily CGK, as any other publication.
Daniel, I know nothing about AI or how to “train” an AI but I can imagine that the day will come when one can give an AI a prior art starting point and an objective technical problem, and ask the AI what hints or suggestions it can find in the prior art, how to solve the problem. I await the day when EPO Examiners do this, as a matter of course, as an early step in their enquiry into obviousness.
If the EESR, at the outset of the EPO prosecution, presents Applicant with the answer from the AI, that Applicant will then be well-placed immediately to explain to the Examiner, concisely and persuasively, how its claimed solution was indeed not obvious.
In this way, it might well be possible (don’t you think) for the EPO to manage with far fewer Examiners than it has today.
Dear Max Drei,
It is certainly the hope of the upper management of the EPO that examiners can be replaced to a large extent by AI. May be the day might come when one can give an AI a prior art starting point and an objective technical problem. I think this day is still quite far away from us.
According to R 42(1,b), the description should indicate the background art, as far as is known to the applicant. Presently this is generally done by giving some bibliographic reference like a patent number.
Without giving a proper bibliographic reference, the problem starts for an AI. When you speak about an objective technical problem, you need to have already carried out a search in order to determine the CPA. This is a second hurdle for an AI. At best it can therefore be the problem the applicant wanted to solve.
Roughly an AI is a black box from which you get an output after it has been trained on a large amount of data, and you have given an input different from the training data. Accepting an answer from an AI without knowing how the answer was generated, is something which, as a matter of principle, will be difficult to accept for applicants. At least the legal framework would need to be adapted should the result of AI ought to be accepted by applicants.
In view of the variety of technical domains you might need at least one AI per IPC sub-class. For this reason, I am not even sure if my grandchildren will see something like this, but I accept to be proved wrong.