SIGIR-AP 2023 Tutorial:
Recent Advances in Generative Information Retrieval

1CAS Key Lab of Network Data Science and Technology, ICT, CAS, University of Chinese Academy of Sciences, 2University of Amsterdam

Sunday November 26 13:00 - 16:30 (GMT+8) @ Room A

About this tutorial

Generative retrieval (GR) has become a highly active area of information retrieval (IR) that has witnessed significant growth recently. Compared to the traditional ``index-retrieve-then-rank'' pipeline, the GR paradigm aims to consolidate all information within a corpus into a single model. Typically, a sequence-to-sequence model is trained to directly map a query to its relevant document identifiers (i.e., docids). This tutorial offers an introduction to the core concepts of the GR paradigm and a comprehensive overview of recent advances in its foundations and applications.

We start by providing preliminary information covering foundational aspects and problem formulations of GR. Then, our focus shifts towards recent progress in docid design, training approaches, inference strategies, and the applications of GR. We end by outlining remaining challenges and issuing a call for future GR research. This tutorial is intended to be beneficial to both researchers and industry practitioners interested in developing novel GR solutions or applying them in real-world scenarios.

Schedule

Our tutorial is scheduled for November 26th from 13:00 to 16:30 (GMT+8). Please note that there could be revisions to the presentation slides. [Slides]

Time Section Presenter
13:00 — 13:10 Section 1: Introduction Maarten de Rijke
13:10 — 13:30 Section 2: Definition & Preliminaries Jiafeng Guo
13:30 — 14:30 Section 3: Docid designs Yubao Tang
14:30 — 14:45 15min coffee break
14:45 — 15:20 Section 4: Training approaches Ruqing Zhang
15:20 — 15:40 Section 5: Inference strategies Ruqing Zhang
15:40 — 16:00 Section 6: Applications Yubao Tang
16:00 — 16:10 Section 7: Challenges & Opportunities Maarten de Rijke
16:10 — 16:30 Q & A All

Reading List

The tutorial extensively covers papers highlighted in bold.


Section 3: Docid design

3.1 Pre-defined docids

3.1.1 A single docid represents a document
3.1.1.1 Number-based docids

Unstructured atomic integers


Naively structured strings


Semantically structured strings


Product quantization strings


3.1.1.2 Word-based docids

Titles


URLs


Pseudo queries


Important terms


3.1.2 Multiple docids represent a document

3.1.2.1 Single type


3.1.2.2 Diverse types

3.2 Learnable docids



Section 4: Training approaches

4.1 Stationary scenarios

4.1.1 Supervised learning

4.1.2 Pre-training

4.1.3 Listwise optimization

4.2 Dynamic scenarios



Section 5: Inference strategies

5.1 A single docid represents a document

Constrained beam search with prefix tree


Constrained greedy search with inverted index


5.2 Multiple docids represent a document

Constrained beam search with FM-index


Aggregation functions



Section 6: Applications

6.1 Knowledge-intensive language tasks (KILT)


6.2 Multi-hop retrieval


6.3 Recommendation


6.4 Code retrieval


BibTeX

@inproceedings{tang-2023-recent,
      author = {Tang, Yubao and Zhang, Ruqing and Guo, Jiafeng and de Rijke, Maarten},
      booktitle = {SIGIR-AP 2023: 1st International ACM SIGIR Conference on Information Retrieval in the Asia Pacific},
      date-added = {2023-10-07 17:24:48 +0200},
      date-modified = {2023-10-07 17:26:24 +0200},
      month = {November},
      publisher = {ACM},
      title = {Recent Advances in Generative Information Retrieval},
      year = {2023}
}