search

yingtaoluo / Spatial-Temporal-Attention-Network-for-POI-Recommendation

Codes for a WWW'21 Paper. POI recommender system for location/trajectory prediction.
https://doi.org/10.1145/3442381.3449998
164 stars 39 forks source link

Can you please explain your trajectory encoding process? #3

Closed princeagarwal12 closed 3 years ago

princeagarwal12 commented 3 years ago

Hello! Thanks for sharing the dataset and the resources. Can you please explain a bit in detail your data encoding method? Some of the approaches have created the location embeddings using skip-gram like approaches. I can see that you've used transition matrices and checkin sequences and then created embeddings but I am unable to understand what you are doing after that and how you're feeding it to attention laters? It would be great if you could please explain this a bit.

yingtaoluo commented 3 years ago

Hi, Prince.

Thanks for asking this. Pre-training of embedding is an effective approach and can further improve the performance for sure. Unfortunately, the focus and contribution of this paper are not on embedding pre-training, and pretraining is not used in baselines, so we do not use it in our paper. Nevertheless, it will be a contribution if you conceive new ideas to improve embedding efficiency.

Best, Yingtao

On Thu, Mar 18, 2021 at 2:29 PM Prince Agarwal @.***> wrote:

Hello! Thanks for sharing the dataset and the resources. Can you please explain a bit in detail your data encoding method? Some of the approaches have created the location embeddings using skip-gram like approaches. I can see that you've used transition matrices and checkin sequences and then created embeddings but I am unable to understand what you are doing after that and how you're feeding it to attention laters? It would be great if you could please explain this a bit.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/yingtaoluo/Spatial-Temporal-Attention-Network-for-POI-Recommendation/issues/3, or unsubscribe https://github.com/notifications/unsubscribe-auth/AH5PX4YOFMDA4W3ITLYO7ALTEGMTLANCNFSM4ZL7UPUA .

princeagarwal12 commented 3 years ago

Thanks Yingtao, for your reply.

ZQSong1997 commented 3 years ago

谢谢，请允许我直接用中文提问吧。我其实知道怎么样从txt转换为csv，这很简单。 我真正疑惑的是因为我在你们的代码里面没发现如何将从网站下载得到的txt原始数据处理为csv这一步骤。所以我不确定是否可以直接先将Gowalla_edges. txt 和gowalla_totalcheckers.txt这样的文件改为csv后，再直接让load.py模块进行处理。但我感觉这种方式是不行的，所以想请问你们是如何将原始的txt格式的dataset变成所需要的npy/csv文件？ 简单的说，就是想知道你们如何将Gowalla_edges. txt和gowalla_totalcheckers.txt处理为gowalla_POI.csv和gowalla.csv这两个代码中所依赖的文件？

yingtaoluo commented 3 years ago

Hi,

Like I said before, please ask the corresponding author for this. I am not in charge of the preprocessing. This is collaboration work. It is perfectly fine to transform txt to csv. You only need to define the DataFrame you need (which we have defined in the problem description section) and discard irrelevant features. I am not sure I follow. This should not be a problem to my best knowledge. If you have other questions, feel free to email me.

Best, Yingtao

On Sat, Apr 10, 2021 at 1:42 AM Zhouchen @.***> wrote:

谢谢，请允许我直接用中文提问吧。我其实知道怎么样从txt转换为csv，这很简单。 我真正疑惑的是因为我在你们的代码里面没发现如何将从网站下载得到的txt原始数据处理为csv这一步骤。所以我不确定是否可以直接先将Gowalla_edges. txt 和gowalla_totalcheckers.txt这样的文件改为csv后，再直接让load.py模块进行处理。但我感觉这种方式是不行的，所以想请问你们是如何将原始的txt格式的dataset变成所需要的npy/csv文件？ 简单的说，就是想知道你们如何将Gowalla_edges. txt和gowalla_totalcheckers.txt处理为gowalla_POI.csv和gowalla.csv这两个代码中所依赖的文件？

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/yingtaoluo/Spatial-Temporal-Attention-Network-for-POI-Recommendation/issues/3#issuecomment-816846641, or unsubscribe https://github.com/notifications/unsubscribe-auth/AH5PX47LMTNJH5EPRFDUF3DTH4367ANCNFSM4ZL7UPUA .