Christophe Van Gysel

[Manos Tsagkias @samanos]

Happy to share the preprint for our #sigir2020 short paper "Predicting Entity Popularity to Improve Spoken Entity Recognition by Virtual Assistants" w/ the Siri Speech team at @Apple – https://t.co/ck4C6zTRWc

[Maarten de Rijke @mdr]

Our CIKM 2018 paper Mix ‘n Match: Integrating Text Matching and Product Substitutability within Product Search is online now. See https://t.co/0J2PELLEEp @cvangysel @ekanou

[Christophe Van Gysel @cvangysel]

Full paper “Mix 'n Match: Integrating Text Matching and Product Substitutability within Product Search”, work I did at @ILPS_Amsterdam w/ @mdr and @ekanou, accepted at #CIKM2018!

[Christophe Van Gysel @cvangysel]

https://t.co/VHxKsRQfWY #SIGIR2018

[Christophe Van Gysel @cvangysel]

Come check out our #SIGIR18 poster for your daily recommended dose of IR evaluation measures! w/ @mdr

[Christophe Van Gysel @cvangysel]

@AdamOptimizer I.e., use the external corpus to extract semantic relations amongst words and then manually construct NVSM vectors from the overlapping terms in your retrieval collection and your external corpus. I haven't tried this though, so I don't know how well this would work.

[Christophe Van Gysel @cvangysel]

@AdamOptimizer I guess you could always get embeddings trained on another corpus, look for the closest in-collection word vectors for out-of-collection terms and then assume that each out-of-collection term is some kind of combination of the NVSM vectors of the in-collection terms.

[Christophe Van Gysel @cvangysel]

@AdamOptimizer I'll fix the error and upload a new version to arXiv in a few days. Have you tried our implementation (https://t.co/ER5RgYVnCr) rather than implementing it yourself?

[Christophe Van Gysel @cvangysel]

@AdamOptimizer Yep, seems like it. Good catch!

[Innovation Center for Artificial Intelligence @ic4ai]

Hello world! The Innovation Center for Artificial Intelligence https://t.co/WNhNm6UXhf

[Christophe Van Gysel @cvangysel]

Our short paper “Pytrec_eval: An Extremely Fast Python Interface to trec_eval” has been accepted at #SIGIR2018 w/ @mdr. Code: https://t.co/VHxKsRQfWY

[Christophe Van Gysel @cvangysel]

Happy to release cuNVSM (https://t.co/ER5RgYVnCr): the C++/#CUDA open-source release that accompanies our #TOIS paper (https://t.co/zXd5awWIMY) on Neural Vector Space Models for #IR! Comes with end-to-end tutorial (https://t.co/BjffM09mC9) showing how ranking signal can be used.

[Bhaskar Mitra | ভাস্কর মিত্র @UnderdogGeek]

@cvangysel is next talking about semantic matching. #NN4IR #ECIR2018

[Christophe Van Gysel @cvangysel]

.@TomKenter talking about distributed representations at #NN4IR #ECIR2018

[Christophe Van Gysel @cvangysel]

Slides of the #NN4IR tutorial at #ECIR2018 are now online at https://t.co/ZwKkEDJQ4x! Come join us at the "Titane 2" room!

@TomKenter @UnderdogGeek @mdr

[Christophe Van Gysel @cvangysel]

Schedule of the #NN4IR tutorial at #ECIR2018 is now online: https://t.co/ZwKkEDJQ4x. See you on Monday in Grenoble! w/ @TomKenter @mdr @UnderdogGeek

[Christophe Van Gysel @cvangysel]

@pommedeterre33 @mdr @ekanou Working on it right now.

[Christophe Van Gysel @cvangysel]

Our paper "Neural Vector Spaces for Unsupervised Information Retrieval" has been accepted for publication at ACM #TOIS (preprint: https://t.co/zXd5awWIMY) w/ @mdr and @ekanou

[Christophe Van Gysel @cvangysel]

Animation showing how #NVSM (https://t.co/zXd5awWIMY) learns #unsupervised semantic document #representations at different stages in the training process (iteration 0 to 15) on the Reuters-21578 collection. Colours represent document topics. #search #AI #machinelearning