Deep Learning Relevance: Creating Relevant Information (as Opposed to Retrieving it)

Christina Lioma; Birger Larsen; Casper Petersen; Jakob Grue Simonsen

Deep Learning Relevance: Creating Relevant Information (as Opposed to Retrieving it)

Christina Lioma, Birger Larsen, Casper Petersen, Jakob Grue Simonsen

Research output: Contribution to book/anthology/report/conference proceeding › Article in proceeding › Research › peer-review

Abstract

What if Information Retrieval (IR) systems did not just retrieve relevant information that is stored in their indices, but could also "understand" it and synthesise it into a single document? We present a preliminary study that makes a first step towards answering this question. Given a query, we train a Recurrent Neural Network (RNN) on existing relevant information to that query. We then use the RNN to "deep learn" a single, synthetic, and we assume, relevant document for that query. We design a crowdsourcing experiment to assess how relevant the "deep learned" document is, compared to existing relevant documents. Users are shown a query and four wordclouds (of three existing relevant documents and our deep learned synthetic document). The synthetic document is ranked on average most relevant of all.

Original language	English
Title of host publication	Neu-IR '16 SIGIR Workshop on Neural Information Retrieval, July 21, 2016, Pisa, Italy
Publication date	2016
Publication status	Published - 2016
Event	Neu-IR '16 SIGIR Workshop on Neural Information Retrieval, July 21, 2016, Pisa, Italy - Pisa, Italy Duration: 21 Jul 2016 → 21 Jul 2016

Workshop

Workshop	Neu-IR '16 SIGIR Workshop on Neural Information Retrieval, July 21, 2016, Pisa, Italy
Country/Territory	Italy
City	Pisa
Period	21/07/2016 → 21/07/2016

Access to Document

https://arxiv.org/abs/1606.07660Licence: Unspecified

AUB Link

Search for the material in Aalborg University Library's search engine

Cite this

@inproceedings{4e5697dcac3e42558c934afe42773b79,

title = "Deep Learning Relevance: Creating Relevant Information (as Opposed to Retrieving it)",

abstract = "What if Information Retrieval (IR) systems did not just retrieve relevant information that is stored in their indices, but could also {"}understand{"} it and synthesise it into a single document? We present a preliminary study that makes a first step towards answering this question. Given a query, we train a Recurrent Neural Network (RNN) on existing relevant information to that query. We then use the RNN to {"}deep learn{"} a single, synthetic, and we assume, relevant document for that query. We design a crowdsourcing experiment to assess how relevant the {"}deep learned{"} document is, compared to existing relevant documents. Users are shown a query and four wordclouds (of three existing relevant documents and our deep learned synthetic document). The synthetic document is ranked on average most relevant of all.",

author = "Christina Lioma and Birger Larsen and Casper Petersen and Simonsen, {Jakob Grue}",

year = "2016",

language = "English",

booktitle = "Neu-IR '16 SIGIR Workshop on Neural Information Retrieval, July 21, 2016, Pisa, Italy",

note = "Neu-IR '16 SIGIR Workshop on Neural Information Retrieval, July 21, 2016, Pisa, Italy, Neu-IR ; Conference date: 21-07-2016 Through 21-07-2016",

}

Lioma, C, Larsen, B, Petersen, C & Simonsen, JG 2016, Deep Learning Relevance: Creating Relevant Information (as Opposed to Retrieving it). in Neu-IR '16 SIGIR Workshop on Neural Information Retrieval, July 21, 2016, Pisa, Italy. Neu-IR '16 SIGIR Workshop on Neural Information Retrieval, July 21, 2016, Pisa, Italy, Pisa, Italy, 21/07/2016. <https://arxiv.org/abs/1606.07660>

TY - GEN

T1 - Deep Learning Relevance

T2 - Neu-IR '16 SIGIR Workshop on Neural Information Retrieval, July 21, 2016, Pisa, Italy

AU - Lioma, Christina

AU - Larsen, Birger

AU - Petersen, Casper

AU - Simonsen, Jakob Grue

PY - 2016

Y1 - 2016

N2 - What if Information Retrieval (IR) systems did not just retrieve relevant information that is stored in their indices, but could also "understand" it and synthesise it into a single document? We present a preliminary study that makes a first step towards answering this question. Given a query, we train a Recurrent Neural Network (RNN) on existing relevant information to that query. We then use the RNN to "deep learn" a single, synthetic, and we assume, relevant document for that query. We design a crowdsourcing experiment to assess how relevant the "deep learned" document is, compared to existing relevant documents. Users are shown a query and four wordclouds (of three existing relevant documents and our deep learned synthetic document). The synthetic document is ranked on average most relevant of all.

AB - What if Information Retrieval (IR) systems did not just retrieve relevant information that is stored in their indices, but could also "understand" it and synthesise it into a single document? We present a preliminary study that makes a first step towards answering this question. Given a query, we train a Recurrent Neural Network (RNN) on existing relevant information to that query. We then use the RNN to "deep learn" a single, synthetic, and we assume, relevant document for that query. We design a crowdsourcing experiment to assess how relevant the "deep learned" document is, compared to existing relevant documents. Users are shown a query and four wordclouds (of three existing relevant documents and our deep learned synthetic document). The synthetic document is ranked on average most relevant of all.

M3 - Article in proceeding

BT - Neu-IR '16 SIGIR Workshop on Neural Information Retrieval, July 21, 2016, Pisa, Italy

Y2 - 21 July 2016 through 21 July 2016

ER -

Deep Learning Relevance: Creating Relevant Information (as Opposed to Retrieving it)

Abstract

Workshop

Access to Document

AUB Link

Fingerprint

Cite this