The ability to perform multi-modal multi-hop reasoning by iteratively integrating information across various modalities and external knowledge is critical for addressing complex real-world challenges. However, existing Multi-modal Large Language Models (MLLMs) are predominantly limited to single-step reasoning, as existing benchmarks lack the complexity needed to evaluate and drive multi-hop abilities. To bridge this gap, we introduce MMhops, a novel, large-scale benchmark designed to systematically evaluate and foster multi-modal multi-hop reasoning. MMhops dataset comprises two challenging task formats, Bridging and Comparison, which necessitate that models dynamically construct complex reasoning chains by integrating external knowledge. To tackle the challenges posed by MMhops, we propose MMhops-R1, a novel multi-modal Retrieval-Augmented Generation (mRAG) framework for dynamic reasoning. Our framework utilizes reinforcement learning to optimize the model for autonomously planning reasoning paths, formulating targeted queries, and synthesizing multi-level information. Comprehensive experiments demonstrate that MMhops-R1 significantly outperforms strong baselines on MMhops, highlighting that dynamic planning and multi-modal knowledge integration are crucial for complex reasoning. Moreover, MMhops-R1 demonstrates strong generalization to tasks requiring fixed-hop reasoning, underscoring the robustness of our dynamic planning approach.

Large language models (LLMs) are trained on vast amounts of text from the Internet, but do they truly understand the viral content that rapidly spreads online-commonly known as memes?In this paper, we introduce CHIME, a dataset for CHinese Internet Meme Explanation.The dataset comprises popular phrase-based memes from the Chinese Internet, annotated with detailed information on their meaning, origin, example sentences, types, etc.To evaluate whether LLMs understand these memes, we designed two tasks.In the first task, we assessed the models' ability to explain a given meme, identify its origin, and generate appropriate example sentences.The results show that while LLMs can explain the meanings of some memes, their performance declines significantly for culturally and linguistically nuanced meme types.Additionally, they consistently struggle to provide accurate origins for the memes.In the second task, we created a set of multiple-choice questions (MCQs) requiring LLMs to select the most appropriate meme to fill in a blank within a contextual sentence.While the evaluated models were able to provide correct answers, their performance remains noticeably below human levels.We have made CHIME public 1 and hope it will facilitate future research on computational meme understanding.

Recent advances in Large Multi-modal Models (LMMs) have demonstrated their remarkable success as general-purpose multi-modal assistants, with particular focuses on holistic image- and video-language understanding. Conversely, less attention has been given to scaling fine-grained pixel-level understanding capabilities, where the models are expected to realize pixel-level alignment between visual signals and language semantics. Some previous studies have applied LMMs to related tasks such as region-level captioning and referring expression segmentation. However, these models are limited to performing either referring or segmentation tasks independently and fail to integrate these fine-grained perception capabilities into visual reasoning. To bridge this gap, we propose UniPixel, a large multi-modal model capable of flexibly comprehending visual prompt inputs and generating mask-grounded responses. Our model distinguishes itself by seamlessly integrating pixel-level perception with general visual understanding capabilities. Specifically, UniPixel processes visual prompts and generates relevant masks on demand, and performs subsequent reasoning conditioning on these intermediate pointers during inference, thereby enabling fine-grained pixel-level reasoning. The effectiveness of our approach has been verified on 10 benchmarks across a diverse set of tasks, including pixel-level referring/segmentation and object-centric understanding in images/videos. A novel PixelQA task that jointly requires referring, segmentation, and question answering is also designed to verify the flexibility of our method.

Dominant dual-encoder models enable efficient image-text retrieval but suffer from limited accuracy, while the cross-encoder models offer higher accuracy at the expense of efficiency. Distilling cross-modality matching knowledge from cross-encoder to dual-encoder provides a natural approach to harness their strengths. Thus, we investigate the following valuable question: how to make cross-encoder a good teacher for dual-encoder? Our findings are threefold: (1) Cross-modal similarity score distribution of cross-encoder is more concentrated, while the result of dual-encoder is nearly normal, making vanilla logit distillation less effective. However, ranking distillation remains practical, as it is not affected by the score distribution. (2) Only the relative order between hard negatives conveys valid knowledge, while the order information between easy negatives has little significance. (3) Maintaining the coordination between distillation loss and dual-encoder training loss is beneficial for knowledge transfer. Based on these findings, we propose a novel Contrastive Partial Ranking Distillation (CPRD) method, which implements the objective of mimicking relative order between hard negative samples with contrastive learning. This approach coordinates with the training of the dual-encoder, effectively transferring valid knowledge from the cross-encoder to the dual-encoder. Extensive experiments on image-text retrieval and ranking tasks show that our method surpasses other distillation methods and significantly improves the accuracy of dual-encoder.

Recent advances in Video Large Language Models (Video-LLMs) have demonstrated their great potential in general-purpose video understanding. To verify the significance of these models, a number of benchmarks have been proposed to diagnose their capabilities in different scenarios. However, existing benchmarks merely evaluate models through video-level question-answering, lacking fine-grained event-level assessment and task diversity. To fill this gap, we introduce E.T. Bench (Event-Level & Time-Sensitive Video Understanding Benchmark), a large-scale and high-quality benchmark for open-ended event-level video understanding. Categorized within a 3-level task taxonomy, E.T. Bench encompasses 7.3K samples under 12 tasks with 7K videos (251.4h total length) under 8 domains, providing comprehensive evaluations. We extensively evaluated 8 Image-LLMs and 12 Video-LLMs on our benchmark, and the results reveal that state-of-the-art models for coarse-level (video-level) understanding struggle to solve our fine-grained tasks, e.g., grounding event-of-interests within videos, largely due to the short video context length, improper time representations, and lack of multi-event training data. Focusing on these issues, we further propose a strong baseline model, E.T. Chat, together with an instruction-tuning dataset E.T. Instruct 164K tailored for fine-grained event-level understanding. Our simple but effective solution demonstrates superior performance in multiple scenarios.

Previous work for video captioning aims to objectively describe the video content but the captions lack human interest and attractiveness, limiting its practical application scenarios. The intention of video title generation (video titling) is to produce attractive titles, but there is a lack of benchmarks. This work offers CREATE, the first large-scale Chinese shoRt vidEo retrievAl and Title gEneration dataset, to assist research and applications in video titling, video captioning, and video retrieval in Chinese. CREATE comprises a high-quality labeled 210 K dataset and two web-scale 3 M and 10 M pre-training datasets, covering 51 categories, 50K+ tags, 537K+ manually annotated titles and captions, and 10M+ short videos with original video information. This work presents ACTEr, a unique Attractiveness-Consensus-based Title Evaluation, to objectively evaluate the quality of video title generation. This metric measures the semantic correlation between the candidate (model-generated title) and references (manual-labeled titles) and introduces attractive consensus weights to assess the attractiveness and relevance of the video title. Accordingly, this work proposes a novel multi-modal ALignment WIth Generation model, ALWIG, as one strong baseline to aid future model development. With the help of a tag-driven video-text alignment module and a GPT-based generation module, this model achieves video titling, captioning, and retrieval simultaneously. We believe that the release of the CREATE dataset, ACTEr metric, and ALWIG model will encourage in-depth research on the analysis and creation of Chinese short videos.

Dominant pre-training works for image-text retrieval adopt “dual-encoder” architecture to enable high efficiency, where two encoders are used to extract image and text representations and contrastive learning is employed for global alignment. However, coarse-grained global alignment ignores detailed semantic associations between image and text. In this work, we propose a novel proxy task, named Visual-Language Error Modeling (ViLEM), to inject detailed image-text association into “dual-encoder” model by “proofreading” each word in the text against the corresponding image. Specifically, we first edit the image-paired text to automatically generate diverse plausible negative texts with pre-trained language models. ViLEM then enforces the model to discriminate the correctness of each word in the plausible negative texts and further correct the wrong words via resorting to image information. Further-more, we propose a multi-granularity interaction framework to perform ViLEM via interacting text features with both global and local image features, which associates local text semantics with both high-level visual context and multi-level local visual information. Our method surpasses state-of-the-art “dual-encoder” methods by a large margin on the image-text retrieval task and significantly improves discriminativeness to local textual semantics. Our model can also generalize well to video-text retrieval.

Video Tagging intends to infer multiple tags spanning relevant content for a given video. Typically, video tags are freely defined and uploaded by a variety of users, so they have two characteristics: abundant in quantity and disordered intra-video. It is difficult for the existing multilabel classification and generation methods to adapt directly to this task. This paper proposes a novel generative model, Order-Prompted Tag Sequence Generation (OP-TSG), according to the above characteristics. It regards video tagging as a tag sequence generation problem guided by sample-dependent order prompts. These prompts are semantically aligned with tags and enable to decouple tag generation order, making the model focus on modeling the tag dependencies. Moreover, the word-based generation strategy enables the model to generate novel tags. To verify the effectiveness and generalization of the proposed method, a Chinese video tagging benchmark CREATE-tagging, and an English image tagging benchmark Pexel-tagging are established. Extensive results show that OP-TSG is significantly superior to other methods, especially the results on rare tags improve by 3.3% and 3% over SOTA methods on CREATE-tagging and Pexel-tagging, and novel tags generated on CREATE-tagging exhibit a tag gain of 7.04%.

Previous dual-encoder pre-training methods for video-text retrieval employ contrastive learning for cross-modal alignment in a latent space. However, such learned latent spaces often result in modality gap problem [26]. In this paper, we introduce a novel SemVTR framework designed to learn semantics-grounded video-text representations in a vocabulary space, in which each dimension corresponds to a semantic concept represented by a word. The representation is obtained by grounding video and text into semantically-related dimensions with high activation values. As video-text pairs share grounded dimensions, their vocabulary representations are expected to cluster together and thus alleviate modality gap problem. So, the crux of our method lies in grounding video and text into vocabulary space. Specifically, we propose a Multi-Granularity Video Semantics Grounding approach and a Textual Semantics Preserving training strategy. The visualization illustrates that SemVTR obtains semantics-gronded vocabulary representation and also alleviates the modality gap problem. SemVTR significantly outperforms existing methods on four video-text retrieval benchmarks.

Open- vocabulary object detection aims to detect novel object categories beyond the training set. The advanced open- vocabulary two-stage detectors employ instance-level visual-to- visual knowledge distillation to align the visual space of the detector with the semantic space of the Pre-trained Visual-Language Model (PVLM). However, in the more efficient one-stage detector, the absence of class-agnostic object proposals hinders the knowledge distil-lation on unseen objects, leading to severe performance degradation. In this paper, we propose a hierarchical visual-language knowledge distillation method, i.e., Hi-erKD, for open-vocabulary one-stage detection. Specifi-cally, a global-level knowledge distillation is explored to transfer the knowledge of unseen categories from the PVLM to the detector. Moreover, we combine the proposed global-level knowledge distillation and the common instance-level knowledge distillation to learn the knowledge of seen and unseen categories simultaneously. Extensive experiments on MS-COCO show that our method significantly surpasses the previous best one-stage detector with 11.9% and 6.7% AP50 gains under the zero-shot detection and generalized zero-shot detection settings, and reduces the AP50performance gap from 14% to 7.3% compared to the best two-stage detector. Code will be released at this url11https://qithub.com/menqqiDyanqqe/HierKD.

One important necessity is to filter out the irrelevant information and organize the relevant information into meaningful categories. However, developing text classifiers often requires a large number of labeled documents as training examples. Manually labeling documents is costly and time-consuming. More importantly, it becomes unrealistic to know all the categories covered by the documents beforehand. Recently, a few methods have been proposed to label documents by using a small set of relevant keywords for each category, known as dataless text classification . In this article, we propose a seed-guided topic model for the dataless text filtering and classification (named DFC). Given a collection of unlabeled documents, and for each specified category a small set of seed words that are relevant to the semantic meaning of the category, DFC filters out the irrelevant documents and classifies the relevant documents into the corresponding categories through topic influence. DFC models two kinds of topics: category-topics and general-topics . Also, there are two kinds of category-topics: relevant-topics and irrelevant-topics. Each relevant-topic is associated with one specific category, representing its semantic meaning. The irrelevant-topics represent the semantics of the unknown categories covered by the document collection. And the general-topics capture the global semantic information. DFC assumes that each document is associated with a single category-topic and a mixture of general-topics. A novelty of the model is that DFC learns the topics by exploiting the explicit word co-occurrence patterns between the seed words and regular words (i.e., non-seed words) in the document collection. A document is then filtered, or classified, based on its posterior category-topic assignment. Experiments on two widely used datasets show that DFC consistently outperforms the state-of-the-art dataless text classifiers for both classification with filtering and classification without filtering. In many tasks, DFC can also achieve comparable or even better classification accuracy than the state-of-the-art supervised learning solutions. Our experimental results further show that DFC is insensitive to the tuning parameters. Moreover, we conduct a thorough study about the impact of seed words for existing dataless text classification techniques. The results reveal that it is not using more seed words but the document coverage of the seed words for the corresponding category that affects the dataless classification performance.

Many applications require semantic understanding of short texts, and inferring discriminative and coherent latent topics is a critical and fundamental task in these applications. Conventional topic models largely rely on word co-occurrences to derive topics from a collection of documents. However, due to the length of each document, short texts are much more sparse in terms of word co-occurrences. Recent studies show that the Dirichlet Multinomial Mixture (DMM) model is effective for topic inference over short texts by assuming that each piece of short text is generated by a single topic. However, DMM has two main limitations. First, even though it seems reasonable to assume that each short text has only one topic because of its shortness, the definition of “shortness” is subjective and the length of the short texts is dataset dependent. That is, the single-topic assumption may be too strong for some datasets. To address this limitation, we propose to model the topic number as a Poisson distribution, allowing each short text to be associated with a small number of topics (e.g., one to three topics). This model is named PDMM. Second, DMM (and also PDMM) does not have access to background knowledge (e.g., semantic relations between words) when modeling short texts. When a human being interprets a piece of short text, the understanding is not solely based on its content words, but also their semantic relations. Recent advances in word embeddings offer effective learning of word semantic relations from a large corpus. Such auxiliary word embeddings enable us to address the second limitation. To this end, we propose to promote the semantically related words under the same topic during the sampling process, by using the generalized Pólya urn (GPU) model. Through the GPU model, background knowledge about word semantic relations learned from millions of external documents can be easily exploited to improve topic modeling for short texts. By directly extending the PDMM model with the GPU model, we propose two more effective topic models for short texts, named GPU-DMM and GPU-PDMM. Through extensive experiments on two real-world short text collections in two languages, we demonstrate that PDMM achieves better topic representations than state-of-the-art models, measured by topic coherence. The learned topic representation leads to better accuracy in a text classification task, as an indirect evaluation. Both GPU-DMM and GPU-PDMM further improve topic coherence and text classification accuracy. GPU-PDMM outperforms GPU-DMM at the price of higher computational costs.

Developing text classifiers often requires a large number of labeled documents as training examples. However, manually labeling documents is costly and time-consuming. Recently, a few methods have been proposed to label documents by using a small set of relevant keywords for each category, known as dataless text classification. In this paper, we propose a Seed-Guided Topic Model (named STM) for the dataless text classification task. Given a collection of unlabeled documents, and for each category a small set of seed words that are relevant to the semantic meaning of the category, the STM predicts the category labels of the documents through topic influence. STM models two kinds of topics: category-topics and general-topics. Each category-topic is associated with one specific category, representing its semantic meaning. The general-topics capture the global semantic information underlying the whole document collection. STM assumes that each document is associated with a single category-topic and a mixture of general-topics. A novelty of the model is that STM learns the topics by exploiting the explicit word co-occurrence patterns between the seed words and regular words (i.e., non-seed words) in the document collection. A document is then labeled, or classified, based on its posterior category-topic assignment. Experiments on two widely used datasets show that STM consistently outperforms the state-of-the-art dataless text classifiers. In some tasks, STM can also achieve comparable or even better classification accuracy than the state-of-the-art supervised learning solutions. Our experimental results further show that STM is insensitive to the tuning parameters. Stable performance with little variation can be achieved in a broad range of parameter settings, making it a desired choice for real applications.

For many applications that require semantic understanding of short texts, inferring discriminative and coherent latent topics from short texts is a critical and fundamental task. Conventional topic models largely rely on word co-occurrences to derive topics from a collection of documents. However, due to the length of each document, short texts are much more sparse in terms of word co-occurrences. Data sparsity therefore becomes a bottleneck for conventional topic models to achieve good results on short texts. On the other hand, when a human being interprets a piece of short text, the understanding is not solely based on its content words, but also her background knowledge (e.g., semantically related words). The recent advances in word embedding offer effective learning of word semantic relations from a large corpus. Exploiting such auxiliary word embeddings to enrich topic modeling for short texts is the main focus of this paper. To this end, we propose a simple, fast, and effective topic model for short texts, named GPU-DMM. Based on the Dirichlet Multinomial Mixture (DMM) model, GPU-DMM promotes the semantically related words under the same topic during the sampling process by using the generalized Polya urn (GPU) model. In this sense, the background knowledge about word semantic relatedness learned from millions of external documents can be easily exploited to improve topic modeling for short texts. Through extensive experiments on two real-world short text collections in two languages, we show that GPU-DMM achieves comparable or better topic representations than state-of-the-art models, measured by topic coherence. The learned topic representation leads to the best accuracy in text classification task, which is used as an indirect evaluation.

Stack Overflow and MedHelp are examples of domain-specific community-based question answering (CQA) systems. Different from CQA systems for general topics (e.g., Yahoo! Answers, Baidu Knows), questions and answers in domain-specific CQA systems are mostly in the same topical domain, enabling more comprehensive interaction between users on fine-grained topics. In such systems, users are more likely to ask questions on unfamiliar topics and to answer questions matching their expertise. Users can also vote answers based on their judgements. In this paper, we propose a Tri-Role Topic Model (TRTM) to model the tri-roles of users (i.e., as askers, answerers, and voters, respectively) and the activities of each role including composing question, selecting question to answer, contributing and voting answers. The proposed model can be used to enhance CQA systems from many perspectives. As a case study, we conducted experiments on ranking answers for questions on Stack Overflow, a CQA system for professional and enthusiast programmers. Experimental results show that TRTM is effective in facilitating users getting ideal rankings of answers, particularly for new and less popular questions. Evaluated on nDCG, TRTM outperforms state-of-the-art methods.

Micro-blogging services and location-based social networks, such as Twitter, Weibo, and Foursquare, enable users to post short messages with timestamps and geographical annotations. The rich spatial-temporal-semantic information of individuals embedded in these geo-annotated short messages provides exciting opportunity to develop many context-aware applications in ubiquitous computing environments. Example applications include contextual recommendation and contextual search. To obtain accurate recommendations and most relevant search results, it is important to capture users’ contextual information (e.g., time and location) and to understand users’ topical interests and intentions. While time and location can be readily captured by smartphones, understanding user’s interests and intentions calls for effective methods in modeling user mobility behavior. Here, user mobility refers to who visits which place at what time for what activity . That is, user mobility behavior modeling must consider user (Who), spatial (Where), temporal (When), and activity (What) aspects. Unfortunately, no previous studies on user mobility behavior modeling have considered all of the four aspects jointly, which have complex interdependencies. In our preliminary study, we propose the first solution named W 4 (short for Who, Where, When, and What) to discover user mobility behavior from the four aspects. In this article, we further enhance W 4 and propose a nonparametric Bayesian model named EW 4 (short for Enhanced W 4 ). EW 4 requires no parameter tuning and achieves better results over W 4 in our experiments. Given some of the four aspects of a user (e.g., time), our model is able to infer information of the other aspects (e.g., location and topical words). Thus, our model has a variety of context-aware applications, particularly in contextual search and recommendation. Experimental results on two real-world datasets show that the proposed model is effective in discovering users’ spatial-temporal topics. The model also significantly outperforms state-of-the-art baselines for various tasks including location prediction for tweets and requirement-aware location recommendation.

The adoption of hashtags in major social networks including Twitter, Facebook, and Google+ is a strong evidence of its importance in facilitating information diffusion and social chatting. To understand the factors (e.g., user interest, posting time and tweet content) that may affect hashtag annotation in Twitter and to capture the implicit relations between latent topics in tweets and their corresponding hashtags, we propose two PLSA-style topic models to model the hashtag annotation behavior in Twitter. Content-Pivoted Model (CPM) assumes that tweet content guides the generation of hashtags while Hashtag-Pivoted Model (HPM) assumes that hashtags guide the generation of tweet content. Both models jointly incorporate user, time, hashtag and tweet content in a probabilistic framework. The PLSA-style models also enable us to verify the impact of social factor on hashtag annotation by introducing social network regularization in the two models. We evaluate the proposed models using perplexity and demonstrate their effectiveness in two applications: retrospective hashtag annotation and related hashtag discovery. Our results show that HPM outperforms CPM by perplexity and both user and time are important factors that affect model performance. In addition, incorporating social network regularization does not improve model performance. Our experimental results also demonstrate the effectiveness of our models in both applications compared with baseline methods.

Micro-blogging services, such as Twitter, and location-based social network applications have generated short text messages associated with geographic information, posting time, and user ids. The availability of such data received from users offers a good opportunity to study the user's spatial-temporal behavior and preference. In this paper, we propose a probabilistic model W4 (short for Who+Where+When+What) to exploit such data to discover individual users' mobility behaviors from spatial, temporal and activity aspects. To the best of our knowledge, our work offers the first solution to jointly model individual user's mobility behavior from the three aspects. Our model has a variety of applications, such as user profiling and location prediction; it can be employed to answer questions such as ``Can we infer the location of a user given a tweet posted by the user and the posting time?" Experimental results on two real-world datasets show that the proposed model is effective in discovering users' spatial-temporal topics, and outperforms state-of-the-art baselines significantly for the task of location prediction for tweets.

The availability of user check-in data in large volume from the rapid growing location based social networks (LBSNs) enables many important location-aware services to users. Point-of-interest (POI) recommendation is one of such services, which is to recommend places where users have not visited before. Several techniques have been recently proposed for the recommendation service. However, no existing work has considered the temporal information for POI recommendations in LBSNs. We believe that time plays an important role in POI recommendations because most users tend to visit different places at different time in a day, \eg visiting a restaurant at noon and visiting a bar at night. In this paper, we define a new problem, namely, the time-aware POI recommendation, to recommend POIs for a given user at a specified time in a day. To solve the problem, we develop a collaborative recommendation model that is able to incorporate temporal information. Moreover, based on the observation that users tend to visit nearby POIs, we further enhance the recommendation model by considering geographical information. Our experimental results on two real-world datasets show that the proposed approach outperforms the state-of-the-art POI recommendation methods substantially.

Because of Twitter's popularity and the viral nature of information dissemination on Twitter, predicting which Twitter topics will become popular in the near future becomes a task of considerable economic importance. Many Twitter topics are annotated by hashtags. In this article, we propose methods to predict the popularity of new hashtags on Twitter by formulating the problem as a classification task. We use five standard classification models (i.e., Naïve bayes, k‐nearest neighbors, decision trees, support vector machines, and logistic regression) for prediction. The main challenge is the identification of effective features for describing new hashtags. We extract 7 content features from a hashtag string and the collection of tweets containing the hashtag and 11 contextual features from the social graph formed by users who have adopted the hashtag. We conducted experiments on a Twitter data set consisting of 31 million tweets from 2 million Singapore‐based users. The experimental results show that the standard classifiers using the extracted features significantly outperform the baseline methods that do not use these features. Among the five classifiers, the logistic regression model performs the best in terms of the Micro‐F1 measure. We also observe that contextual features are more effective than content features.

Readers of a news article often read its comments contributed by other readers. By reading comments, readers obtain not only complementary information about this news article but also the opinions from other readers. However, the existing ranking mechanisms for comments (e.g., by recency or by user rating) fail to offer an overall picture of topics discussed in comments. In this paper, we first propose to study Topic-driven Reader Comments Summarization (Torcs) problem. We observe that many news articles from a news stream are related to each other; so are their comments. Hence, news articles and their associated comments provide context information for user commenting. To implicitly capture the context information, we propose two topic models to address the Torcs problem, namely, Master-Slave Topic Model (MSTM) and Extended Master-Slave Topic Model (EXTM). Both models treat a news article as a master document and each of its comments as a slave document. MSTM model constrains that the topics discussed in comments have to be derived from the commenting news article. On the other hand, EXTM model allows generating words of comments using both the topics derived from the commenting news article, and the topics derived from all comments themselves. Both models are used to group comments into topic clusters. We then use two ranking mechanisms Maximal Marginal Relevance (MMR) and Rating & Length (RL) to select a few most representative comments from each comment cluster. To evaluate the two models, we conducted experiments on 1005 Yahoo! News articles with more than one million comments. Our experimental results show that EXTM significantly outperforms MSTM by perplexity. Through a user study, we also confirm that the comment summary generated by EXTM achieves better intra-cluster topic cohesion and inter-cluster topic diversity.

Hashtags are widely used in Twitter to define a shared context for events or topics. In this paper, we aim to predict hashtag popularity in near future (i.e., next day). Given a hashtag that has the potential to be popular in the next day, we construct a hashtag profile using the tweets containing the hashtag, and extract both content and context features for hashtag popularity prediction. We model this prediction problem as a classification problem and evaluate the effectiveness of the extracted features and classification models.