This decision concerns a system for controlling and optimizing information distribution between users in an information exchange. However, since the distinguishing features were considered non-technical, the EPO refused to grant a patent. Here are the practical takeaways from the decision T 1369/20 (Controlling information flow/MCFADDEN) of October 27, 2023 of the Technical Board of Appeal 3.5.01.
Key takeaways
Whether or not to send a particular message via an electronic communication network is not necessarily a technical decision.
The invention
The Board in charge summarized the subject-matter of the application underlying the present decision as follows:
2.1 The invention in claim 1 of this request concerns a method for determining whether to include an information item into an information stream in an information exchange. This could be, for example, a football score in a news feed, but, of course, the claim covers many other possibilities. The news item comprises contents, for example “Manchester City 2 – Nottingham Forest 0”, and a meta description comprising hierarchical data used to describe the content (for example, “Football” – “English Football” – “English Premier League”). The information exchange is a portal or service that facilitates the flow of information items to a consumer.
2.2 Claim 1 refers to a “participation metric” which is a measure of information item consumption or interaction with the information item by the consumer. In the oral proceedings, the appellant explained that this was based on historical data about previous items delivered to the consumer. What news articles were consumed? Did the consumer click on the link? Was there some other action indicating that they had read the story?
Further down in claim 1, there is a “participation prediction mapping” process which relates an “expected item value” to a “predicted participation level”. According to the appellant, this means that there is a statistical relation connecting the metadata with participation, for example a ratio – the number of times the consumer had the chance to interact with a category (English football scores) to the times they actually interacted with it. In other words, the “item value” in claim 1 is a statistical relation between the metadata and the participation metric, and it is used to predict the participation level of an information item that has not been seen yet. It could be seen as a score. If the consumer likes English football, the metadata “English football” will have a high item value. This means that the predicted participation of an unseen English football score (having metadata “English football” with a high item value) will be high.
The method further comprises obtaining a distribution of information items over a two dimensional range of priorities. This can be explained with reference to Figure 8. The distribution is a frequency distribution of the number of items per time unit for each point on that graph. Each item has an item value (priority) between -1 and 1 for the user and, though not in the claim, also an item value for the producer. Those values will be used to place the items in the grid in Figure 8. For example, “English football” has a high item value for the consumer and also a high value for the producer and will be placed in the upper right corner of the graph.
The next step in the method is to, for a region of the distribution of information items:
(a) compute a number of information items over the region of the distribution,
(b) compute a specified expected item value over the region of the distribution,
(c) evaluate the predicted participation level from the participation prediction mapping for the specified expected item value to determine where the predicted participation level for the specified item value is approximately equal to the number of information items in the region or where the predicted participation level from the participation prediction map is less than the number of items; and
select an include region of the distribution as the region where the predicted participation level for the specified item value is approximately equal to the number of information items in the region.
In other words, the algorithm is looking for a region where the predicted participation for the items of the region is approximately equal to the number of items in the region. The aim is to choose a number of items that matches the user’s participation. If it is predicted that the user is going to consume 11 items, the algorithm chooses a region with 11 items.
These item are included in the information stream that is delivered to the consumer.
Fig. 1 of WO 2014/145365 A2
-
Claim 1 (Third Auxiliary Request)
Is it technical?
With regard to technical character, the Board came to following conclusion:
3.1 […] claim 1 implies some sort of computer system, and, therefore, the claimed subject-matter has technical character overall, and is not excluded under Article 52(2) and (3) EPC.
With regards to inventive step, the Board cited T 641/00 – Two identities/COMVIK and mentioned that the question is, consequently, to what extent the claimed method provides a technical effect. The Board further discussed:
4.2 The examining division considered that the claimed method related to data analysis with the goal of optimising information distribution. This was considered to be a mental act or a non-technical method of organising information. No technical effect could be identified. The motivation behind the invention did not constitute a technical problem as it did not involve any technical skills. Thus, starting from a notoriously known conventional networked computer as the “closest prior art”, the technical problem to be solved was how to implement the non-technical features on the known computer system. The implementation at the level of detail of claim 1, would, however, have been obvious for the skilled person.
In response, the appellant argued that:
4.3 […] the invention increased the efficiency of the information exchange system, because items that were not consumed were not sent, and, at the same time, the consumer was not given too few information items. This was a technical effect. Moreover, the control of the information flow took place automatically, without active involvement of the user. This was also technical. Furthermore, the regulation of information flow was triggered by bandwidth limitations and nothing else. Thus, the invention was based on technical considerations.
The Board was not convinced by the appellant’s arguments but rather agrees with the examining division. More specifically:
[…] the decision whether or not to include the information item into the information stream in claim 1 is not based on any technical parameters of the data network or computer system. It is rather based on the content of the message in view of the customer’s interests and priorities. The method gives the consumer as many information items as he is likely to consume. That is not a technical optimisation. It is an information optimisation.
It is clear that any message that is sent via an electronic communication network has an effect on network traffic. However, that does not mean that the decision whether or not to send a particular message is necessarily a technical one. There has to be a further technical effect going beyond the normal and inevitable effects of sending (or not sending) a message.
The same is true for the automation aspect. This is an effect of using a computer, which was known and obvious at the priority date.
As a result, the Board found that claim 1 lacks an inventive step and dismissed the appeal.
More information
You can read the whole decision here: T 1369/20 (Controlling information flow/MCFADDEN) of October 27, 2023.