Abstract:
Recommendations based on visual resemblance play a significant role in e-
commerce platforms since they assist customers in selecting the desired items
more quickly. Despite its huge potential, the number of recommendation
systems on this topic is limited. An effective recommendation system is
necessary to properly sort, order, and communicate relevant product material
or in- formation to users. Here in our research, we presented an efficient E-
commerce
similar
product
network
model
for
visually
similar
recommendations. To achieve our objective, we have performed image feature
extraction and generating embeddings(Denoising-Autoencoder) using deep
learning techniques and built an Index tree using the K-Nearest Neighbour
classification(KNN) algorithm. Further, we have fetched top-N the near
similar items using a distance measure. We have benchmarked our model in
terms of accuracy and error rate, and it’s a new approach with 93% accuracy.
