Maybe you are looking for CroQS: Cross-modal Query Suggestion for Text-to-Image Retrieval

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper formalizes cross-modal query suggestion as a task, contributes the CroQS benchmark of 50 queries and 295 human-validated clusters, and shows that captioning and LLM baselines far outperform the initial query on cluster…

desk verdict A genuinely useful benchmark for a new task, with an evaluation that leans too hard on CLIP and an abstract that overstates the baselines' gains; worth reviewing after tightening. read the letter →

arxiv 2412.13834 v1 pith:KM4X2FRT submitted 2024-12-18 cs.IR cs.AIcs.LG

Giacomo Pacini , Fabio Carrara , Nicola Messina , Nicola Tonellotto , Giuseppe Amato , Fabrizio Falchi This is my paper · ORCID

classification cs.IRcs.AIcs.LG

keywords cross-modalquerysuggestiontext-to-imageretrievalreformulationbenchmarkimagegroupcaptioningCLIPlargelanguagemodelsinformation

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper establishes cross-modal query suggestion as a measurable retrieval task: starting from a text query and the images it retrieves, a system should propose several refined textual queries, each aimed at one visually coherent subset of the results. To make the task testable, the paper builds CroQS, a benchmark of 50 initial queries with 295 human-validated image clusters and a human-written suggested query per cluster. It defines three evaluation axes—cluster specificity, representativeness, and similarity to the original query—and adapts image-captioning and LLM-based baselines to the task. On the benchmark, those baselines improve cluster-specificity recall by more than 115% and representativeness mean average precision by more than 52% relative to the initial query, while remaining below human-annotated suggestions. If the benchmark holds up, it gives interactive text-to-image search a standard way to compare query-suggestion methods.

What carries the argument

The load-bearing object is the CroQS benchmark itself: 50 initial queries built on the COCO train split, with 295 human-validated clusters (5.9 clusters per query on average) and one human-annotated suggested query per cluster. The evaluation uses three metrics: RecallCluster, the fraction of a cluster's images ranked in the top $k$ of the initial result set when re-ranked by the suggested query; representativeness measures (Recall, NDCG, MAP) computed by retrieving the cluster's images from the whole collection with the suggested query; and similarity to $q_0$ via Jaccard and CLIP textual similarity. The baseline machinery is prototype captioning—averaging or selecting the most representative CLIP image embedding of a cluster and decoding it with ClipCap or DeCap, optionally conditioned on $q_0$—and GroupCap, which captions the most representative images and asks an LLM to summarize them together with $q_0$.

What would settle it

Take a held-out set of CroQS clusters and have human annotators rank candidate suggested queries by how well each captures the cluster; compare those rankings with the CLIP-similarity ordering used for the metrics. A large disagreement would show that the benchmark's numbers measure the embedding model rather than the quality of the suggestions.

Watch

Extended reading notes

Core claim

The central claim is that query suggestion can be moved from text-only retrieval to cross-modal retrieval, and that the right unit of evaluation is the semantic cluster: a good suggested query should pick out one coherent group of images inside the initial result set without drifting far from the user's original wording. CroQS supplies the ground truth for this claim by fixing the clusters through human judgment, so that different methods can be compared purely on the queries they generate. The paper further claims that off-the-shelf captioning models applied to cluster prototypes, and LLMs that summarize per-image captions, both produce useful suggestions: the captioning-derived methods are the most cluster-specific, the LLM method is the most balanced and stays closest to the initial query, and both beat simply re-using the initial query by large margins. Human annotations still score highest on specificity and representativeness, which the paper reads as evidence that the task is well posed but not yet solved.

Load-bearing premise

The evaluation assumes that CLIP's text-image similarity matches the way humans group images; CLIP is used both to form the groups and to score the suggestions, so a mismatch would make the reported gains partly an artifact of the system agreeing with itself.

Editorial extensions

If this is right

CroQS can serve as a reusable testbed: any future query-suggestion model can be scored on the same fixed clusters and human-annotated queries without redoing the annotation effort.
Because the benchmark fixes clusters, the comparison isolates query generation from clustering, so method improvements can be attributed to the suggestion module rather than to different groupings.
The gains over the initial query show that even simple captioning and LLM baselines make interactive exploration of image collections feasible in practice.
The gap to human performance identifies a concrete target: methods that keep the specificity of captioning models while matching the query fidelity of LLMs.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Beyond the paper, a natural extension is a user study: the benchmark's CLIP-based metrics could be checked against whether human searchers actually prefer the suggested queries when browsing.
Because the clusters are defined in CLIP space, the benchmark's scores are partly a measure of CLIP self-consistency; re-annotating or re-clustering with a different embedding would reveal how much of the reported gain is tied to that specific representation.
The same task definition could transfer to video retrieval or large image archives where captions are absent, since the method only needs embeddings and a captioning model.
CroQS could be extended to a setting where each cluster has multiple equally valid suggested queries, letting metrics reward diversity of good queries rather than a single reference.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Useful paper. The core contribution is the benchmark. CroQS gives 50 queries, 295 human-curated clusters, and human-annotated suggested queries. That is a real resource for a task that currently has no standard evaluation, and it is reusable even if you never use their baselines. The task formalization—conditioning on an initial query and clusters of unannotated images—is a genuine gap relative to visual query suggestion and group captioning. The paper also ships the dataset, code, and notebooks, which is the right way to do this.

The baselines are straightforward adaptations, not novel methods, and the paper is honest about that. Tables are internally consistent, standard deviations are reported, and the human upper-bound comparison is useful.

Now the soft spots. The stress-test concern is real but needs to be stated fairly. CLIP appears in cluster construction, prototype selection, and the metrics. Because clusters were created with CLIP features, evaluating suggestions by CLIP retrieval can reward text that hovers near the CLIP centroid even if it would not strike a user as the best label. Human annotations define the groups, but they are not used to score the metrics. So the headline gains over q0 are partly a CLIP self-consistency result. This does not sink the benchmark—it is a construct-validity gap. An easy fix is to re-score the human and baseline suggestions with a different embedding or with human relevance judgments on a subset. Until that check exists, the quantitative conclusions should be read cautiously.

The abstract overstates the results. More than 115% RecallCluster is true for several baselines, but only GroupCap clears 52% representativeness mAP; DeCap and ClipCap are around 33%. And representativeness Recall for every baseline is below q0's 0.52. The trade-off is plausible—specificity vs. recall—but the abstract should say which method and which metric.

Smaller issues: Eq. 4 does not include the top-100 cutoff mentioned in the text, NDCG@10 is stated but not defined, GroupCap's prompt and number of captioned images are not given, and there is no inter-annotator agreement for the human clusters or suggestions. These are fixable in revision.

Who is this for? Anyone working on interactive text-to-image retrieval, query reformulation, or benchmark design. It deserves a serious referee. I would send it out, ask for the metric definitions and the CLIP validation, and accept after those are addressed.

Referee Report

4 major / 5 minor

Summary. The paper formalizes a new task, cross-modal query suggestion for text-to-image retrieval, in which a system takes an initial textual query q0 and the visual result set R(q0, I), partitions it into semantic clusters, and generates one suggested query per cluster. The authors contribute CroQS, a benchmark of 50 initial queries with 295 human-validated clusters and human-annotated reference suggestions built on COCO train images. They define three evaluation properties (cluster specificity, representativeness, and similarity to the original query), propose corresponding metrics, and adapt two families of baselines: prototype-captioning methods (ClipCap, DeCap, ClipCapq0) and an LLM-based caption-summarization method (GroupCap). Their experiments report that the baselines improve cluster-specificity recall and representativeness mAP over the initial query, while remaining below human performance on most metrics. The dataset, baseline code, and notebooks are publicly released.

Significance. If the evaluation is valid, the paper makes a useful contribution: it defines a new interactive-retrieval task, provides a reusable human-annotated benchmark, and supplies transparent baselines with standard deviations and open resources. The benchmark itself is potentially valuable independently of the headline numbers, since it provides a shared protocol for future work. However, the central quantitative claims rest on CLIP-based metrics that are entangled with the same model used to build clusters and drive the baselines. Until that construct-validity gap is addressed, the reported improvements over q0 should be treated as provisional. The abstract's paired improvement claim is also not fully supported by the tables as written. These issues are fixable with additional validation experiments and a more careful statement of results, so the paper merits a major revision rather than rejection.

major comments (4)

[§3.2, §4.2 (Eqs. 3–5), §5.1] CLIP appears in three places in the evaluation loop: it produces the initial result set and the clusters in §3.2, it supplies the prototype points and representative-image selection for the baselines in §5.1 (Eqs. 5–6), and it scores cluster specificity and representativeness in §4.2 (Eqs. 3–4). Because clusters are subsets of a CLIP-based ranking, a suggested query that moves CLIP rankings toward the CLIP cluster centroid will receive high scores even if it does not align with human semantic grouping; the human-annotated reference suggestions in CroQS are scored with the same CLIP metrics rather than used as an external check. The headline gains in the Abstract (>115% RecallCluster and >52% mAP) may therefore partly reflect within-model self-consistency. Please add a validation study using an independent image-text embedding model and/or human ratings on a sample of clusters, and report the resulting metric values or agreement scores, or explicitly qualify the quantitative conclusions.
[Abstract, Table 1] The Abstract states that both LLM-based and captioning-based methods improve cluster-specificity recall by more than 115% and representativeness mAP by more than 52% with respect to the initial query. Table 1 shows that no adapted method achieves both improvements simultaneously: ClipCap and DeCap improve RecallCluster by roughly +184% and +179% respectively but mAP by only about +33%, while GroupCap improves mAP by about +52% but RecallCluster by about +105%. Please rephrase the claim so that the metric improvements are attributed to the specific configurations that achieve them, or present paired results that support the current wording. As written, the central quantitative claim in the Abstract is not supported by the reported tables.
[§4.2, Eq. (4); §5.2] Eq. (4) defines Recall as |Ci ∩ R(q̂_i, I)| / |Ci| over the whole collection, but §5.2 states that representativeness scores are computed on the top-100 documents. The equation needs an explicit truncation parameter, e.g., R_{100}(q̂_i, I); otherwise the metric as defined differs from the metric that is reported, and for any cluster with |Ci| > 100 the denominator must be reconciled with the cutoff. Please also clarify whether Eq. (3)'s retrieval set R(q0, I) is truncated and how the top-k cutoff k=|Ci| interacts with that truncation. This is not purely notational: the reported representativeness numbers depend on the cutoff.
[§5.1, Figure 2b, Table 3] GroupCap's configuration is incompletely specified: the number k of representative images whose captions are fed to the LLM, the number of few-shot examples in the prompt, the exact prompt template, and the sampling/decoding parameters for Mistral-7B and Llama3-8B are not reported. Without these details, the baseline scores in Tables 1 and 3 cannot be reproduced from the paper text. Please provide a full configuration in an appendix or in the released code documentation, and state which settings were used for the results reported in each table.

minor comments (5)

[§4.2, Tables 1–3] The 'CLIP Query' column is not defined in the metric section; please state that it is the cosine similarity between the CLIP text embeddings of q0 and q̂_i, and specify which CLIP model variant and text encoder (e.g., ViT-B/32, which layer) was used.
[§4.2, Eqs. (3)–(4)] The notation R(q0, I) is used both for the initial result set and for the ranked list produced by a suggested query; please introduce separate notation for a ranked list and for its top-k truncation, as the current overloading makes the formulas harder to follow.
[§5.3, §6] The sentence in §5.3 that GroupCapLLama3 obtains a 'slightly better macro-averaged score than the human-annotated suggestions' refers only to the similarity-to-q0 property, while §6 says the methods are 'relatively far from human performance'. Please add a qualifier so that the two statements do not appear contradictory.
[Table 3] The label 'LLama30.90' appears to be a typesetting error; the writing should be 'GroupCapLLama3' with the score properly separated. The text also alternates between 'Llama3' and 'LLama3'; please standardize the capitalization.
[§4.1 and project page] The paper should state the license and terms of use for the released benchmark, baseline code, and notebooks, and should clarify that COCO's license permits redistribution of the derived cluster/query annotations.

Circularity Check

1 steps flagged · score 3.0 of 10

Cluster construction, prototype-captioning baselines, and the main metrics all use CLIP, so the reported gains are partly a within-model round trip; human-annotated ground truth prevents full circularity.

self definitional [Sec. 3.2 (cluster construction), Sec. 5.1 and Eq. 5 (prototype captioning), Sec. 4.2 Eq. 3 (metric)]
"We delegate the partitioning of the initial result set R(q0, I) into different semantic groups {C1, C2, . . . , CM} to off-the-shelf clustering algorithms operating on the semantic image representations of the collection (i.e., CLIP visual features in our experiments, the same adopted for the initial cross-modal search) ... The core idea is to select a prototype point that is close enough to each image in the group ... This prototype is then decoded to generate a representative text for the image group. ... RecallCluster = |Ci ∩ R[:k](q̂i, R(q0, I))|/k"

For the captioning baselines, q̂i is generated by decoding a CLIP-space prototype of Ci (Eq. 5), while Eq. 3 scores q̂i by re-ranking R(q0, I) with a CLIP text-image search. Since the clusters themselves were formed from CLIP visual features of the same result set, a suggested query is rewarded for moving toward the CLIP centroid of its own cluster; the reported >115% RecallCluster gain over q0 is therefore substantially a CLIP self-consistency score rather than an independent measure of human semantic grouping. The same CLIP geometry also underlies the representativeness metrics in Eq. 4, which rank the whole collection. Human cluster validation and imperfect text decoding prevent a full tautology, but the headline quantitative claim is not externally validated.

full rationale

The paper contains no fitted-parameter prediction and no load-bearing self-citation: the baselines are standard captioning/LLM models, and the human-annotated CroQS clusters and reference suggestions provide independent ground truth for the benchmark. The only circularity concern is evaluative: CLIP is used to form the initial clusters (Sec. 3.2), to define the CLIP-space prototypes from which the captioning baselines decode their suggestions (Sec. 5.1, Eq. 5), and to compute the Cluster Specificity and Representativeness scores (Sec. 4.2, Eqs. 3-4). For prototype-captioning baselines, this makes the specificity score approximately a CLIP round-trip: a suggestion is rewarded for retrieving the same CLIP cluster that generated it. The >115% and >52% improvements over q0 are therefore partly a within-model consistency result. I do not rate this higher because human judgment was used to validate clusters and write reference suggestions, and because a non-CLIP method could in principle be evaluated on the same benchmark; the limitation is construct validity rather than a derivation that equals its inputs by construction.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

No fitted parameters appear in the core contribution; the benchmark is built from human annotation and CLIP-based clustering. The main assumptions are that CLIP similarity corresponds to semantic relevance and that single-annotator human queries are a valid gold standard. Metric thresholds and GroupCap's representative image count are hand choices that the numerical results depend on.

free parameters (3)

GroupCap representative image count = not reported
Number of cluster images caption-coded before LLM summarization; affects GroupCap output and is not specified in Sec 5.1.
Evaluation cutoffs (top-100, NDCG@10, k=|Ci|) = top-100 for representativeness, NDCG@10, k=|Ci| for RecallCluster
Hand-chosen retrieval cutoffs in Sec 5.2; the reported numbers depend on them, especially the 'top 100 documents' wording.
Prototype selection for captioning baselines = centroid (Eq 5) preferred over representative (Eq 6)
Ablation choice in Sec 5.3; method comparisons in Table 1 use the best configuration, which is not fully specified in equations.

assumptions (5)

domain assumption CLIP is a valid semantic space for measuring query-image relevance.
Used for initial retrieval, clustering (Sec 3.2), prototype captioning (Sec 5.1), and all evaluation metrics (Sec 4.2).
domain assumption Human-annotated suggested queries are reliable ground truth.
Single annotator, no agreement statistics, annotations used as reference performance (Sec 4.1).
domain assumption COCO train split is an appropriate image collection for the benchmark.
CroQS built on COCO train images (Sec 4.1), no justification why this collection represents general image search.
domain assumption Off-the-shelf clustering (CLIP features) yields coherent semantic groups.
Task formulation delegates grouping to clustering (Sec 3.2); CroQS fixes clusters manually to remove variability, but validity of the cluster concept depends on CLIP.
standard math Standard retrieval metrics measure the defined properties.
RecallCluster, Recall, NDCG, and MAP are conventional; their adequacy for specificity and representativeness is asserted in Sec 4.2.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Maybe you are looking for CroQS: Cross-modal Query Suggestion for Text-to-Image Retrieval." pith.science (2026). https://pith.science/paper/KM4X2FRT

@misc{pith2026241213834,
  author       = {Pith},
  title        = {Pith review of: Maybe you are looking for CroQS: Cross-modal Query Suggestion for Text-to-Image Retrieval},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KM4X2FRT}},
  note         = {Machine review of arXiv:2412.13834}
}

read the original abstract

Query suggestion, a technique widely adopted in information retrieval, enhances system interactivity and the browsing experience of document collections. In cross-modal retrieval, many works have focused on retrieving relevant items from natural language queries, while few have explored query suggestion solutions. In this work, we address query suggestion in cross-modal retrieval, introducing a novel task that focuses on suggesting minimal textual modifications needed to explore visually consistent subsets of the collection, following the premise of ''Maybe you are looking for''. To facilitate the evaluation and development of methods, we present a tailored benchmark named CroQS. This dataset comprises initial queries, grouped result sets, and human-defined suggested queries for each group. We establish dedicated metrics to rigorously evaluate the performance of various methods on this task, measuring representativeness, cluster specificity, and similarity of the suggested queries to the original ones. Baseline methods from related fields, such as image captioning and content summarization, are adapted for this task to provide reference performance scores. Although relatively far from human performance, our experiments reveal that both LLM-based and captioning-based methods achieve competitive results on CroQS, improving the recall on cluster specificity by more than 115% and representativeness mAP by more than 52% with respect to the initial query. The dataset, the implementation of the baseline methods and the notebooks containing our experiments are available here: https://paciosoft.com/CroQS-benchmark/

Figures

Figures reproduced from arXiv: 2412.13834 by the authors.

**Figure 1.** Cross-modal Query Suggestion. Given an initial query q0 and an image collection I, a cross-modal query suggestion system F returns a set of query suggestions Q based on the visual content of the result set R(q0, I). Ideally, each suggestion qˆi ∈ Q should represent a semantically coherent group Ci ⊂ R(q0, I). The first one is generally automatic and modifies the IR system’s internal representation of the original q… view at source ↗

**Figure 2.** Architectures of the baseline methods proposed. [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗

**Figure 3.** CroQS samples and predictions. Each panel reports a sample of a semantic cluster with its initial query q0, its annotation, and the query suggested by the tested methods. Representative images (Eq. 6) have a colored outline. query suggestion methods. We proposed the CroQS benchmark, composed of a set of 50 diverse queries and a total of 295 image clusters each having a humanannotated query suggestion. We adapted tw… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

29 extracted references · 20 canonical work pages

[1]

In: International Conference on Multimedia Modeling

Amato, G., Bolettieri, P., Carrara, F., Falchi, F., Gennaro, C., Messina, N., Vadicamo, L., Vairo, C.: Visione 5.0: Enhanced user interface and ai models for vbs2024. In: International Conference on Multimedia Modeling. pp. 332–339. Springer (2024)

2024
[2]

arXiv preprint arXiv:2405.19749 (2024)

Bacciu, A., Palumbo, E., Damianou, A., Tonellotto, N., Silvestri, F.: Generating query recommendations via llms. arXiv preprint arXiv:2405.19749 (2024)

arXiv 2024
[3]

In: Proceedings of the ACM on Web Conference 2024

Baek, J., Chandrasekaran, N., Cucerzan, S., Herring, A., Jauhar, S.K.: Knowledge- augmented large language models for personalized contextual query suggestion. In: Proceedings of the ACM on Web Conference 2024. pp. 3355–3366 (2024)

work page 2024
[4]

High Quality Related Search Query Suggestions using Deep Reinforcement Learning

Bodigutla, P.K.: High quality related search query suggestions using deep rein- forcement learning. arXiv preprint arXiv:2108.04452 (2021)

work page Pith review arXiv 2021
[5]

In: Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining

Cao, H., Jiang, D., Pei, J., He, Q., Liao, Z., Chen, E., Li, H.: Context-aware query suggestion by mining click-through and session data. In: Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining. pp. 875–883 (2008)

work page 2008
[6]

arXiv preprint arXiv:2407.21783 (2024)

Dubey, A., Jauhri, A., Pandey, A., Kadian, A., Al-Dahle, A., Letman, A., Mathur, A., Schelten, A., Yang, A., Fan, A., et al.: The llama 3 herd of models. arXiv preprint arXiv:2407.21783 (2024)

arXiv 2024
[7]

In: Proceedings of the 2024 International Conference on Multimedia Retrieval

Gurrin, C., Zhou, L., Healy, G., Bailer, W., Dang Nguyen, D.T., Hodges, S., Jóns- son, B.Þ., Lokoč, J., Rossetto, L., Tran, M.T., et al.: Introduction to the seventh annual lifelog search challenge, lsc’24. In: Proceedings of the 2024 International Conference on Multimedia Retrieval. pp. 1334–1335 (2024)

work page 2024
[8]

Jiang, A.Q., Sablayrolles, A., Mensch, A., Bamford, C., Chaplot, D.S., de las Casas, D., Bressand, F., Lengyel, G., Lample, G., Saulnier, L., Lavaud, L.R., Lachaux, M.A., Stock, P., Scao, T.L., Lavril, T., Wang, T., Lacroix, T., Sayed, W.E.: Mistral 7b (2023), https://arxiv.org/abs/2310.06825

arXiv 2023

Show all 29 references

[9]

In: Proceedings of the 43rd International ACM SIGIR conference on research and development in Information Retrieval

Khattab, O., Zaharia, M.: Colbert: Efficient and effective passage search via con- textualized late interaction over bert. In: Proceedings of the 43rd International ACM SIGIR conference on research and development in Information Retrieval. pp. 39–48 (2020)

2020
[10]

Li, W., Zhu, L., Wen, L., Yang, Y.: Decap: Decoding clip latents for zero-shot captioning via text-only training (2023)

2023
[11]

In: Proceedings of the IEEE/CVF confer- ence on computer vision and pattern recognition

Li, Z., Tran, Q., Mai, L., Lin, Z., Yuille, A.L.: Context-aware group captioning via self-attention and contrastive features. In: Proceedings of the IEEE/CVF confer- ence on computer vision and pattern recognition. pp. 3440–3450 (2020)

2020
[12]

Pacini et al

Lin, T.Y., Maire, M., Belongie, S., Bourdev, L., Girshick, R., Hays, J., Perona, P., Ramanan, D., Zitnick, C.L., Dollár, P.: Microsoft coco: Common objects in context (2014) 14 G. Pacini et al

2014
[13]

Multimedia Systems 29(6), 3481–3504 (2023)

Lokoč, J., Andreadis, S., Bailer, W., Duane, A., Gurrin, C., Ma, Z., Messina, N., Nguyen, T.N., Peška, L., Rossetto, L., et al.: Interactive video retrieval in the age of effective joint embedding deep models: lessons from the 11th vbs. Multimedia Systems 29(6), 3481–3504 (2023)

2023
[14]

Mei,Q.,Zhou,D.,Church,K.:Querysuggestionusinghittingtime.In:Proceedings of the 17th ACM conference on Information and knowledge management. pp. 469– 478 (2008)

2008
[15]

ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM)17(4), 1–23 (2021)

Messina, N., Amato, G., Esuli, A., Falchi, F., Gennaro, C., Marchand-Maillet, S.: Fine-grained visual textual alignment for cross-modal retrieval using transformer encoders. ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM)17(4), 1–23 (2021)

2021
[16]

In: International Conference on Content-based Multimedia Indexing

Messina, N., Stefanini, M., Cornia, M., Baraldi, L., Falchi, F., Amato, G., Cuc- chiara, R.: Aladin: Distilling fine-grained alignment scores for efficient image-text matching and retrieval. In: International Conference on Content-based Multimedia Indexing. pp. 64–70 (2022)

2022
[17]

Mokady, R., Hertz, A., Bermano, A.H.: Clipcap: Clip prefix for image captioning (2021)

2021
[18]

ACM Transactions on Information Systems (TOIS)40(1), 1–27 (2021)

Mustar, A., Lamprier, S., Piwowarski, B.: On the study of transformers for query suggestion. ACM Transactions on Information Systems (TOIS)40(1), 1–27 (2021)

2021
[19]

Ooi, J., Ma, X., Qin, H., Liew, S.C.: A survey of query expansion, query suggestion andqueryrefinementtechniques.In:20154thInternationalConferenceonSoftware Engineering and Computer Systems (ICSECS). pp. 112–117. IEEE (2015)

2015
[20]

IEEE Access (2024)

Phueaksri, I., Kastner, M.A., Kawanishi, Y., Komamizu, T., Ide, I.: Image- collection summarization using scene-graph generation with external knowledge. IEEE Access (2024)

2024
[21]

Radford, A., Kim, J.W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., Krueger, G., Sutskever, I.: Learning transferable visual models from natural language supervision (2021)

2021
[22]

OpenAI blog1(8), 9 (2019)

Radford,A.,Wu,J.,Child,R.,Luan,D.,Amodei,D.,Sutskever,I.,etal.:Language models are unsupervised multitask learners. OpenAI blog1(8), 9 (2019)

2019
[23]

In: The SMART Re- trieval System: Experiments in Automatic Document Processing

Rocchio, J.J.: Relevance feedback in information retrieval. In: The SMART Re- trieval System: Experiments in Automatic Document Processing. Prentice Hall (1971)

1971
[24]

arXiv preprint arXiv:2006.08686 (2020)

Trieu, N., Goodman, S., Narayana, P., Sone, K., Soricut, R.: Multi-image sum- marization: Textual summary from a set of cohesive images. arXiv preprint arXiv:2006.08686 (2020)

2020 arXiv
[25]

In: Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part I 16

Wang, J., Xu, W., Wang, Q., Chan, A.B.: Compare and reweight: Distinctive image captioning using similar images sets. In: Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part I 16. pp. 370–386. Springer (2020)

2020
[26]

In: Proceedings of the 29th ACM International Conference on Multimedia

Wang,J.,Xu,W.,Wang,Q.,Chan,A.B.:Group-baseddistinctiveimagecaptioning with memory attention. In: Proceedings of the 29th ACM International Conference on Multimedia. pp. 5020–5028 (2021)

2021
[27]

IEEE Transactions on Pattern Analysis and Machine Intelligence 45(2), 2088–2103 (2022)

Wang, J., Xu, W., Wang, Q., Chan, A.B.: On distinctive image captioning via comparing and reweighting. IEEE Transactions on Pattern Analysis and Machine Intelligence 45(2), 2088–2103 (2022)

2022
[28]

In: Proceedings of the 2018 World Wide Web Conference

Wu, B., Xiong, C., Sun, M., Liu, Z.: Query suggestion with feedback memory network. In: Proceedings of the 2018 World Wide Web Conference. pp. 1563–1571 (2018) CroQS: Cross-modal Query Suggestion for Text-to-Image Retrieval 15

2018
[29]

In: Proceedings of the 17th ACM international conference on Multimedia

Zha, Z.J., Yang, L., Mei, T., Wang, M., Wang, Z.: Visual query suggestion. In: Proceedings of the 17th ACM international conference on Multimedia. pp. 15–24 (2009)

2009

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[1] [1]

In: International Conference on Multimedia Modeling

Amato, G., Bolettieri, P., Carrara, F., Falchi, F., Gennaro, C., Messina, N., Vadicamo, L., Vairo, C.: Visione 5.0: Enhanced user interface and ai models for vbs2024. In: International Conference on Multimedia Modeling. pp. 332–339. Springer (2024)

2024

[2] [2]

arXiv preprint arXiv:2405.19749 (2024)

Bacciu, A., Palumbo, E., Damianou, A., Tonellotto, N., Silvestri, F.: Generating query recommendations via llms. arXiv preprint arXiv:2405.19749 (2024)

arXiv 2024

[3] [3]

In: Proceedings of the ACM on Web Conference 2024

Baek, J., Chandrasekaran, N., Cucerzan, S., Herring, A., Jauhar, S.K.: Knowledge- augmented large language models for personalized contextual query suggestion. In: Proceedings of the ACM on Web Conference 2024. pp. 3355–3366 (2024)

work page 2024

[4] [4]

High Quality Related Search Query Suggestions using Deep Reinforcement Learning

Bodigutla, P.K.: High quality related search query suggestions using deep rein- forcement learning. arXiv preprint arXiv:2108.04452 (2021)

work page Pith review arXiv 2021

[5] [5]

In: Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining

Cao, H., Jiang, D., Pei, J., He, Q., Liao, Z., Chen, E., Li, H.: Context-aware query suggestion by mining click-through and session data. In: Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining. pp. 875–883 (2008)

work page 2008

[6] [6]

arXiv preprint arXiv:2407.21783 (2024)

Dubey, A., Jauhri, A., Pandey, A., Kadian, A., Al-Dahle, A., Letman, A., Mathur, A., Schelten, A., Yang, A., Fan, A., et al.: The llama 3 herd of models. arXiv preprint arXiv:2407.21783 (2024)

arXiv 2024

[7] [7]

In: Proceedings of the 2024 International Conference on Multimedia Retrieval

Gurrin, C., Zhou, L., Healy, G., Bailer, W., Dang Nguyen, D.T., Hodges, S., Jóns- son, B.Þ., Lokoč, J., Rossetto, L., Tran, M.T., et al.: Introduction to the seventh annual lifelog search challenge, lsc’24. In: Proceedings of the 2024 International Conference on Multimedia Retrieval. pp. 1334–1335 (2024)

work page 2024

[8] [8]

Jiang, A.Q., Sablayrolles, A., Mensch, A., Bamford, C., Chaplot, D.S., de las Casas, D., Bressand, F., Lengyel, G., Lample, G., Saulnier, L., Lavaud, L.R., Lachaux, M.A., Stock, P., Scao, T.L., Lavril, T., Wang, T., Lacroix, T., Sayed, W.E.: Mistral 7b (2023), https://arxiv.org/abs/2310.06825

arXiv 2023