Recom Airbnb embedding

读论文

Real-time Personalization using Embeddings for search Ranking at Airbnb

part1 Introduction

part2 Related work

part3 Methodology

In the following we introduce the proposed methodology for the
task of listing recommendations and listing ranking in search at
Airbnb.
下面我们描述下airbnb在序列推荐任务和搜索序列排序中的方法
We describe two distinct approaches, i.e. listing embeddings
for short-term real-time personalization and user-type & listing
type embeddings for long term personalization, respectively.
我们用了两种不同的方法，一个是对短期兴趣做listing embedding，另一种是对长期兴趣做 usertype&listing type 的embedding

1. Listing Embeddings

2. User-type & Listing-type Embeddings

part4 Experiments

1. Training Listing Embeddings
2. Offline Evaluation of Listing Embeddings
3. Similar Listings using Embeddings
4. Real time personalization in Search Ranking using Embeddings

part5 Conlcusion