The Semantic Entity Retrieval Toolkit (SERT) is a collection of neural entity retrieval algorithms.
Currently, it hosts an implementation of the following models:
- the log-linear model for expertise retrieval, published at WWW 2016
- the latent vector space model for product search, published at CIKM 2016
SERT requires Python 3.5 and assorted modules. The trec_eval utility is required for evaluation and the end-to-end scripts. If you wish to train your models on GPGPUs, you will need a GPU compatible with Theano.
To begin, create a virtual Python environment and install dependencies:
[cvangysel@ilps cvangysel] git clone [email protected]:cvangysel/SERT.git
[cvangysel@ilps cvangysel] cd SERT
[cvangysel@ilps SERT] virtualenv SERT-dev
Using base prefix '/Users/cvangysel/anaconda3'
New python executable in /home/cvangysel/SERT/SERT-dev/bin/python
Installing setuptools, pip, wheel...done.
[cvangysel@ilps SERT] source SERT-dev/bin/activate
(SERT-dev) [cvangysel@ilps SERT] pip install -r requirements.txt
Afterwards, follow the examples for expertise retrieval or product search.
If you use SERT to produce results for your scientific publication, please refer to our WWW 2016 or CIKM 2016 papers:
@inproceedings{VanGysel2016experts,
title={Unsupervised, Efficient and Semantic Expertise Retrieval},
author={Van Gysel, Christophe and de Rijke, Maarten and Worring, Marcel},
booktitle={WWW},
volume={2016},
pages={1069--1079},
year={2016},
organization={The International World Wide Web Conferences Steering Committee}
}
@inproceedings{VanGysel2016products,
title={Learning Latent Vector Spaces for Product Search},
author={Van Gysel, Christophe and de Rijke, Maarten and Kanoulas, Evangelos},
booktitle={CIKM},
volume={2016},
year={2016},
organization={ACM}
}
SERT is licensed under the MIT license. If you modify SERT in any way, please link back to this repository.