1. 1. Introduction

 The Japan Patent Office publishes patent examination cases regarding AI-related technology.

 While AI-related technology is expected to develop in various technological fields, 10 new cases were added to the handbook of patent and utility model as the patent examination cases regarding AI-related technology.

 These cases added as the patent examination cases regarding AI-related technology shall be analyzed in this article.

2. 2. Outline of the added cases

 The added cases (10 in total) break down into 6 cases related to the description requirements (Case Nos. 46 to 51) and 4 cases related to the inventive step (Case Nos. 33 to 36). The added cases cover various technical fields.

〇 : means that the case concerned satisfies the requirement in question.

☓ : means that the case concerned does not satisfies the requirement in question.

△ : means that the case concerned satisfies the requirement in question and the case concerned does not satisfy the requirement in question.
 
 

〇 : means involving inventive step.
☓ : means not involving inventive step.

3. 3. Explanation of the Added Cases

3-1. Cases related to description requirements

 We explain Claim 1 of Cases 46 and 49 as the case of not satisfying the description requirements and Claim 2 of Cases 47 and 49 as the case of satisfying the description requirements.

(1) Case 46

[Claims]

[Claim 1]

 A sugar content estimation system comprising:

 a storage means for storing face images of people and sugar contents of vegetables produced by the people;

 a model generation means for generating a determination model through machine learning, to which a face image of a person is input and from which a sugar content of a vegetable produced by the person is output, using training data containing the face images of the people stored in the storage means and the sugar contents of the vegetables,

 a reception means for receiving an input of an face image; and

 a processing means for outputting, using the generated determination model that has been generated by the model generation means, a sugar content of a vegetable produced by a person that is estimated based on the face image of the person inputted to the reception means.

 
[Overview of the Description]

 It is an object of the present invention to provide a system that estimates a sugar content of a vegetable produced by a person based on his/her face image, taking advantage of the existence of a certain correlation between a face feature of a person and a sugar content of a vegetable produced by the person.

[Overview of Reason for Refusal]

 Article 36(4)(i) (Enablement Requirement)

 According to the description, a human face image is used for an input to a determination model that estimates a sugar content of a vegetable produced by the person. The description says that a face feature is characterized by a head length, face width, nose width, and lip width, for example.

 However, the description only disclose that there is a certain correlation between a face image of a person and a sugar content of a vegetable produced by the person and does not disclose any correlation or the like between them, though disclosing that a face feature is characterized by a head length, face width, nose width, and lip width, for example. It cannot be presumed that there is a correlation or the like between them, even if a common general technical knowledge at the time of filing is taken into consideration. Further, there is no performance evaluation result of an actually generated determination model shown in the description.

(2) Case 47

[Claims]

[Claim1]

 A business plan design apparatus comprising:

 a storage means for storing a stock amount of a specific product;

 a reception means for receiving a web advertisement data and mention data of the specific product;

 a simulation and output means for, using an estimation model that has been trained through machine learning with a training data containing a web advertisement data and mention data of a similar product that has been sold in the past and a sales quantity of the similar product, simulating and outputting a future sales quantity of the specific product estimated based on the web advertisement data and mention data of the specific product;

 a production plan making means for planning a future production quantity of the specific product, based on the stored stock amount and the output sales quantity; and

 an output means for outputting the output sales quantity and the production plan.

[Overview of the Description]

 The business plan design apparatus firstly stores a stock amount of a specific product. The apparatus then obtains an estimated product sales quantity of the product based on an input of a web advertisement data and mention data of the product, using an estimation model that outputs an estimated product sales quantity. In this case, the web advertisement data is the number of times when the specific product publicly appeared on the web. The advertisement includes banner ads, product listing ads, and direct e-mails. The mention data includes reviews on the product or advertisement in web articles, social media, and blogs etc. In the reviews on the product or advertisement, an evaluation value is set so that it becomes greater if there are a lot of positive reviews, and otherwise, it becomes lower. The evaluation value can be obtained through a known computer processing on the text in web articles, social media, and blogs etc.

[Overview of Reason for Refusal]

 There is no reason for refusal found.

[Notes]

 Article 36(4)(i) (Enablement Requirement)

 The description discloses that a web advertisement data and mention data are used. The web advertisement data is based on the number of times when a specific product publicly appeared on the web, and the mention data is based on an evaluation value of reviews on the product or advertisement in web articles, social media, and blogs etc.

 Although the description does not discloses a correlation or the like between the web advertisement data and the mention data, it can be presumed that there is a correlation or the like between them in view of a common general technical knowledge at the time of filing.

(3) Case 49

[Claims]

[Claim 1]

 A body weight estimation system comprising:

 a model generation means for generating an estimation model that estimates a body weight of a person based on a feature value representing a face shape and a body height of the person, through machine learning using training data containing feature values representing face images as well as actual measured values of body heights and body weights of people;

 a reception means for receiving an input of a face image and body height of a person;

 a feature value obtainment means for obtaining a feature value representing a face shape of the person through analysis of the face image of the person that has been received by the reception means; and

 a processing means for outputting an estimated value of a body weight of the person based on the feature value representing the face shape of the person that has been received by the feature value obtainment means and the body height of the person that has been received by the reception means, using the generated estimation model by the model generation means.

[Claim 2]

 The body weight estimation system as in Claim 1, wherein the feature value representing a face shape is a face-outline angle.

[Overview of the Description]

 There is a certain degree of correlation between a face feature and physical size of a person. ..., the inventor found a statistically significant correlation between a cosine of a face-outline angle and BMI (defined as a body weight divided by the square of a body height) of a person. The face-outline angle here means an angle defined between a tangent line to a jaw and a tangent line to a cheek.

 ...This suggests a certain degree of correlation between a body height and weight used for BMI calculation and a face-outline angle. Accordingly, an estimation model with a highly accurate output can be generated through machine learning, using a known machine learning algorithm such as a neural network with a training data. The training data contains actual measured values of face-outline angles, body heights, and body weights. The face-outline angles are obtained through analysis on face images of people.

[Overview of Reason for Refusal]

・Claim 1 : Article 36(6)(i) (Support Requirement)/Article 36(4)(i) (Enablement Requirement)

・Claim 2 : There is no reason for refusal found.

・ Article 36(6)(i) (Support Requirement)/Article 36(4)(i) (Enablement Requirement) : Claim 1

 ..., the description only discloses that any feature value other than a face-outline angle representing a face shape may be obtained from a face image and used. It does not disclose a correlation or the like between (i) a feature value other than a face-outline angle representing a face shape and (ii) a body height, weight, and the like of a person and BMI based on these. Further, it cannot be presumed that there is such a correlation or the like even if a common general technical knowledge at the time of filing is taken into consideration. There is no performance evaluation result disclosed on an estimation model that has actually been generated using a feature value other than a face-outline angle representing a face shape.

[Notes]

Claim 2

 The description discloses that there is a statistically significant correlation between a cosine of a face-outline angle and BMI of a person.

 Based on the disclosure in the description, a person skilled in the art can recognize that there is a certain degree of correlation between a body height and weight and a face-outline angle, and can generate an estimation model using a universal machine learning algorithm with a training data containing actual measured values of body heights, body weights, and face-outline angles.

3-2. Cases related to an inventive step

 We explain Claim 1 of Case 34, and Case 35 as the case of not having an inventive step and Claim 2 of Case 34 as the case of having an inventive step.

(1) Case 34

[Claims]

[Claim 1]

 An estimation system of a hydroelectric power generating capacity of a dam comprising:

 a neural network that is built by means of an information processor, the neural network having an input layer and an output layer, in which an input data to the input layer containing a precipitation amount of the upper stream of a river, a water flow rate of the upper stream of the river, and a water inflow rate into a dam during a predetermined period between a reference time and a predetermined time before the reference time, and an output data from the output layer containing a hydroelectric power generating capacity in the future after the reference time;

 a machine learning unit that trains the neural network using a training data corresponding to actual values of the input data and the output data; and

 an estimation unit that inputs the input data to the neural network that has been trained by the machine learning unit with setting a current time as the reference time, and then calculates an estimated value of a future hydroelectric power generating capacity based on the output data of which reference time is the current time.

[Claim 2]

 The estimation system of a hydroelectric power generating capacity as in Claim 1, wherein the input data to the input layer further contains a temperature of the upper stream of the river during the predetermined period between the reference time and the predetermined time before the reference time.

[Conclusion]

 The invention of Claim 1 does not have an inventive step.

 The invention of Claim 2 has an inventive step.

[Overview of Reason for Refusal]

 The invention of Claim 1 and Cited Invention 1 are different from each other at the point below.

(Difference)

 The invention of Claim 1 realizes an estimation of a hydroelectric power generating capacity by means of a neural network having an input layer and output layer. Meanwhile, Cited Invention 1 realizes an estimation of a hydroelectric power generating capacity by means of a regression equation model.

 The difference is assessed as follows.

 ... Cited Invention 1 and the well-known art are common with each other in estimating a certain output in the future based on an input of time series data in the past, with reference to a correlation among data. Therefore, a person skilled in the art could easily derive a configuration that enables estimation of a hydroelectric power generating capacity, by applying the well-known art to Cited Invention 1 and adopting a trained neural network in substitution of a regression equation model.

(Explanation for no reason for refusal)

 The invention of Claim 2 and Cited Invention 1 are different from each other at the point below.

(Difference)

 The invention of Claim 2 contains, in an input data into an input layer, a temperature of the upperstream of the river during a predetermined period between a reference time and a predetermined time before the reference time. Meanwhile, Cited Invention 1 does not have such a configuration.

 The difference is assessed as follows.

 ... Generally, an input of data of which correlation is unknown may cause a noise in machine learning. However, the invention of Claim 2 uses an input data containing a temperature of the upperstream of the river during a predetermined period between a reference time and a predetermined time before the reference time. This enables a highly accurate estimation of a hydroelectric power generating capacity, taking an increase of inflow rate due to meltwater in the spring into consideration. It is a significant effect that a person skilled in the art cannot expect.

 Accordingly, it does not considered to be a mere workshop modification that can be carried out in application of the well-known art to Cited Invention 1 by a person skilled in the art to contain, in an input data in an estimation of a hydroelectric power generating capacity, a temperature of the upperstream of the river during a predetermined period between a reference time and a predetermined time before the reference time.

 
(2) Case 35

[Claims]

[Claim 1]

 A screw clamping quality estimation apparatus that assesses a screw clamping quality at the time of automatic screw clamping operation by means of a screwdriver comprising:

 a condition measurement unit that measures a set of condition variables containing a rotation speed, angular acceleration, position, and inclination of the screwdriver;

 a machine learning unit that trains a neural network through machine learning by associating, with each other, the set of condition variables measured by the condition measurement unit and the screw clamping quality at the time of automatic screw clamping operation with the use of the set of condition variables; and

 a screw clamping quality estimation unit that estimates a screw clamping quality in response to an input, to the neural network that has been trained by the machine learning unit, of the set of condition variables that have been measured at the time of automatic screw clamping operation by means of a screwdriver.

[Conclusion]

 The invention of Claim 1 does not have an inventive step.

[Overview of Reason for Refusal]

 The invention of Claim 1 and Cited Invention 1 are different with each other at the point below.

(Difference)

 According to the invention of Claim 1, a condition measurement unit measures a set of condition variables containing a rotation speed, angular acceleration, position, and inclination of a screwdriver. Using the set of condition variables containing these four types of variable, a machine learning of a neural network is carried out and a screw clamping quality is estimated. Meanwhile, according to Cited Invention 1, a condition measurement unit measures a set of condition variables containing a rotation speed and angular acceleration of a screwdriver. Using the set of condition variables containing these two types of variable, a machine learning of a neural network is carried out and a screw clamping quality is estimated.

 The difference is assessed as follows.

 Cited Invention 2, in which a screw clamping quality is assessed based on a position and inclination of a screw driver, discloses that there is a correlation between a position and inclination of a screw driver and it affects the assessment. Both Cited Invention 1 and Cited Invention 2 assess a screw clamping quality based on several conditions of a screw driver, and have a common object. Further, it is a common general technical knowledge in the technical field of machine learning to adopt, as an input to a machine learning device, variables that may have a correlation with an output with high possibility, in order to enhance a reliability and accuracy of an output from the machine learning device.

4. 4. Discussion

(1) Regarding description requirements

 In Case 46, since the existence of correlation or the like between various kinds of data included in training data is not supported by the specification or the like and further the existence of some correlation or the like between such data cannot be inferred in light of the common general technical knowledge at the time of filing the application, it is decided that the description requirement is not satisfied.

 In Claim 1 of Case 49, since the existence of correlation or the like between various kinds of data included in training data described in a generic concept is not supported by the specification or the like and further the existence of some correlation or the like between such data cannot be inferred in light of the common general technical knowledge at the time of filing the application, it is decided that the description requirement is not satisfied.

 In Case 47, although any specific correlation or the like between various kinds of data included in training data is not described in the specification or the like, it is decided that the description requirement is satisfied since the existence of correlation or the like between such data can be inferred in light of the common general technical knowledge at the time of filing the application.

 In Claim 2 of Case 49, since the existence of correlation or the like between various kinds of data included in training data is supported by explanation in the specification or the like and statistic information, it is decided that the description requirement is satisfied.

 Therefore, in order to satisfy the description requirements, it is considered to be important that the existence of correlation or the like between various kinds of data included in training data is supported by the specification or the like or the existence of some correlation or the like between such data can be inferred in light of the common general technical knowledge at the time of filing the application.

(2) Regarding an inventive step

 An inventive step of Claim 1 of Case 34 is denied because of a simple change of the estimation means which estimates output data from input data. Also, although detailed explanation was left out in “3-2” above, an inventive step of Case 33 is denied because of a simple systematization of business operations done by human beings by using artificial intelligence.

 Therefore, it is highly likely that an invention step of any invention for application of simple AI would be denied.

 An inventive step of Case 35 is denied since an addition of training data for leaning is a combination of known data so that any prominent effect cannot be recognized. An inventive step of Claim 2 of Case 34 is acknowledged since an addition of training data for learning involves a prominent effect.

 Therefore, in order to acknowledge an inventive step of the invention related to the change of training data, it is considered to be important that a prominent effect is recognized in a change of training data.