update

ST_DM/KDD2021-CHAML/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2021 YudongChen
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

ST_DM/KDD2021-CHAML/README.md

@@ -0,0 +1,92 @@
+## Curriculum Meta-Learning for Next POI Recommendation
+
+The codes for the paper: Curriculum Meta-Learning for Next POI Recommendation (KDD 2021).
+
+This repository is under construction.
+
+### Requirements
+
+```
+python 3.6
+paddlepaddle >= 2.0.0
+numpy >= 1.19.0
+x2paddle (please refer to https://github.com/PaddlePaddle/X2Paddle)
+```
+
+### Run
+
+To run the codes, one should first put the map search data under the path `./data/dataset/`.
+Each city is a `.txt` file, where each line is the search records of one user in that city:
+
+```shell
+Each line:
+{user_id} \t {os_name} \t {search_record1} \t {search_record2} \t ...
+# {os_name}: the operation system of the user's mobile, e.g., "iphone", "andriod", etc.
+Each search record:
+{time}_{poiid}_{poiloc}_{poiname}_{poitype}_{userloc}_{network}
+# {time}: the timestamp when the user searched the POI, e.g., '2021-08-14 10:00:00'
+# {poiid}: the POI ID hashcode of the searched POI, e.g., 'wqxc4t8fd147a'
+# {poiloc}: the Mercator coordinates of the POI, e.g., '(12967374,4810843)'
+# {poiname}: the name of the POI, e.g., 'Happy Cinema'
+# {poitype}: the category of the POI, e.g., 'cinema'
+# {userloc}: the Mercator coordinates of the user when he/she searched the POI
+# {network}: '4G' or 'wifi'
+```
+
+Next, set the config files under the path `./data/config/`:
+
+- `base_cities.txt`: The name of all the base cities (for meta-training). Each line for one city name. The city names **should exactly match** the names in `./data/dataset/`. The same below. E.g.,
+
+  ```
+  beijing
+  shanghai
+  shenzhen
+  ...
+  ```
+
+- `valid_cities.txt`: The name of all the valid cities (for meta-validation).
+
+- `test_cities.txt`: The name of all the test cities (for meta-testing).
+
+- `poi_category.txt`: All the POI categories in the raw dataset. Each line for one POI category. E.g., 
+
+  ```
+  cinema
+  school
+  residential area
+  ...
+  ```
+
+Finally, you can run the following instructions at `./`:
+
+```shell
+sh run_prep.sh
+sh run.sh
+```
+
+As a fake example, you may run as follows to run our codes on fake dataset:
+
+```sh
+python preprocess/fake_data_generator.py
+sh run_prep.sh
+sh run.sh
+# Although the meta-learner cannot learn anything from the fake data..
+```
+
+Notice: there are still some warnings during running the codes, but it's just OK to run.
+
+
+
+### Cite
+
+Please cite the paper: 
+
+```
+@article{Chen2021CurriculumMF,
+  title={Curriculum Meta-Learning for Next POI Recommendation},
+  author={Yudong Chen and Xin Wang and Miao Fan and Jizhou Huang and Shengwen Yang and Wenwu Zhu},
+  journal={Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery \& Data Mining},
+  year={2021}
+}
+```
+

ST_DM/KDD2021-CHAML/config/config-chaml.json

@@ -0,0 +1,25 @@
+{
+    "update_lr": 1e-3,
+    "meta_lr": 1e-3,
+    "update_step": 1,
+    "update_step_test": 1,
+
+    "task_batch_size": 4,
+
+    "train_qry_batch_size": 512,
+    "max_train_steps": 100000,
+    "few_num": 512,
+
+    "num_poi_types": 229,
+    "num_time": 25,
+    "embed_dim": 50,
+    "poiid_dim": 50,
+    "mlp_hidden": 300,
+    "local_fix_var": 1,
+    "global_fix_var": 1,
+
+
+    "sample_batch_size": 1024,
+    "test_task_batch_size": 1,
+    "num_epoch": 10
+}