Similarity-based approach for inventive design solutions assistance

Abstract

With the increasing demand for inventive products, finding out inventive design solutions hidden in different industrial engineering domains has always been a challenge for engineers. In addition, patent documents are full of the latest inventive knowledge inside. In this paper, we rely on the assumption that an engineering problem may have an inventive practical solution in another scientific domain as long as they are similarly described. Therefore, we focus on applying machine learning techniques, more particularly neural networks to determine the similarity between patent problems. Technically, a trained bidirectional LSTM neural network, called Manhattan LSTM is integrated into our approach named SAM-IDM to predict the similarity between sentences. We experimentally show that Manhattan LSTM outperforms other baseline approaches in a labelled sample dataset of SNLI corpus. We then experiment our approach on a real-world U.S. patent dataset and we demonstrate that it presents promising results in terms of sentence similarity matching and inventiveness. An inventive design case is detailed to illustrate its performance and practicality.

Appendices

Appendices

A case studies

We introduce additional real solved use cases found in the literature. We highlight the output of SAM-IDM on these problems. SAM-IDM mines all similar problems and their patents using a fixed similarity threshold. These similar problems and corresponding solutions are investigated by experts to check their potential in creating an inventive solution.

Target Problem 1:“It is thus possible to prevent a decrease in the conductivity of the conductive film 421 due to oxidation.” (US9536627, Physics)

• Similar Problem (Similarity Value: 0.83): “materials function as a heat sink due to the latent heat of vaporization.” (US9537344, Electricity)

• Corresponding Solution: “materials transitioning from a liquid to a gas or vapor, or from a solid to a gas or vapor, could also be used as a heat sink, so long as the temperature associated with the phase change was in the desired range.”

• Latent Inventive Solution Reference: From the corresponding solution according to the similar problem, we notice that Eranna et al. (2004) mentioned oxide materials for development of integrated gas sensors are related to the conductivity.

• Similar Problem (Similarity Value: 0.83): “It may lead to a decrease in the yield in subsequent production steps.” (US9538663, Electricity)

• Corresponding Solution: “connecting a sidewall of the wiring board to a sidewall in an opening of the metal frame by subjecting the metal frame to plastic deformation.”

• Latent Inventive Solution Reference: With the hints of the corresponding solution, we notice that raising the mix ratio of the powdered metal is, in the conductive filler kneading method, able to obtain the desired surface resistance Kawaguchi et al. (2006). This might be useful to solve the target problem.

• Similar Problem (Similarity Value: 0.82): “it is difficult to form a roughened surface.” (US9538642, Electricity)

• Corresponding Solution: “Therefore, in the present embodiment, the above-described thin resin layer 1012 is formed through a method that uses resin contraction caused by heating.”

• Latent Inventive Solution Reference: Based on the corresponding solution, we notice that a solid polymer electrolyte film from hydrogen bonding layer DeLongchamp and Hammond (2004) and a percolating-assisted resin film infusion method Wang et al. (2018) have been proposed to solve the related problem.

Target Problem 2: “Segmentation and analysis can add computational complexity and overhead.” (US9534958, Physics)

• Similar Problem (Similarity Value: 0.86): “which can incur high computational complexity.” (US9538483, Electricity)

• Corresponding Solution: “Some embodiments described herein provide a novel approach to the weighted sum-rate maximization in the MIMO interference network, and apply a novel and efficient algorithm with guaranteed monotonic convergence as well as an elegant way to establish rate duality between an interference network and its reciprocal.”

• Latent Inventive Solution Reference: A Low-Complexity Intrusion Detection Algorithm is proposed by Sajana (2011) to solve the similar target issue.

Target Problem 3: “many of the devices also require the use of a pressure washer.” (US9533320, Performing Operations)

• Similar Problem (Similarity Value: 0.87): “a hole of a microphone is blocked” (US9537985, Electricity)

• Corresponding Solution: “The hall structure 100 according to the present invention can better reduce blocking of a hole of a microphone by combining the hole of the microphone and a hole of a speakerphone or receiver.”

• Latent Inventive Solution Reference: The solution from US9537985 that combining two different holes is close to the target problem’s solution that the device consists of a connector and conduits in US9533320.

Target Problem 4: “Patient-specific implants are expensive to engineer and manufacture. Moreover, the plate can cause bone necrosis if the fit is too snug.” (US9532825, Human Necessities)

• Similar Problem (Similarity Value: 0.81): “reservoirs are expensive and difficult to manufacture. ” (US9534198, Chemistry)

• Corresponding Solution: “One aspect of the present invention is to provide an EC fluid cycling unit that enables fluid level control without the use of expensive ultrasonics or load cells.”

• Latent Inventive Solution Reference: The bioprinted constructs for treatment of degenerated intervertebral disc Costa et al. (2019) that is similar with the fluid cycling unit has been proposed to solve the similar issue with the target problem.