Image flare artifacts, caused by internal light reflections and scattering in camera lenses, degrade image quality by introducing unwanted bright spots, ghosting, and haze. These artifacts distort scene details and negatively impact computer vision applications. To address this, we propose a deep learning model that effectively removes flares from single images while preserving critical visual content. Our method first estimates and refines a depth map of the input image using a Dense Vision Transformer (DPT). The refined depth map is combined with the original RGB image to form a 4 -channel input, which is processed by a U-shaped network. This network uses encoderdecoder blocks to progressively suppress flares, with a novel Hybrid Transformer Enhancement Block (HTEB) at its core to model global relationships and local details. Experiments on the Flare7K++ benchmark and real-world images demonstrate that our approach outperforms existing methods in flare removal accuracy and detail preservation, proving its robustness for practical use. This work underscores the potential of hybrid attention-frequency mechanisms in transformer-based restoration architectures.

Recursion is one of the most important and hardest topics in lower division computer science courses. As it is an advanced programming skill, the best way to learn it is through targeted practice exercises. But the best practice problems are time consuming to manually grade by an instructor. As a consequence, students historically have completed only a small number of recursion programming exercises as part of their coursework. We present a new way for teaching such programming skills. Students view examples and visualizations, then practice a wide variety of automatically assessed, small-scale programming exercises that address the sub-skills required to learn recursion. The basic recursion tutorial (RecurTutor) teaches material typically encountered in CS2 courses. Students who used RecurTutor had significantly better grades on recursion exam questions than did students who used typical instruction. Students who experienced RecurTutor spent significantly more time on solving recursive programming exercises than students who experienced typical instruction, and came out with a significantly higher confidence level.

In this work, we look at problems of urban sensing from the lens of conversations (tweets) on Twitter. Using techniques from statistical natural language processing on geotagged tweets, we identify areas which exhibit similar aggregate behavior, infer the land-use of areas and predict types of individual establishments. We demonstrate our inferences using over two years of Twitter data, for a wide variety of spatial contexts and evaluate our results against existing open data sets. Our results are novel in extremely detailed resolution of their mapping, and demonstrate that tweets can be a very effective urban sensor and in many regards are superior to other data sources for studying urban spaces. Our techniques are language agnostic, and can be applied to any city where enough similar data is available.

Microblogs, such as Twitter, are a way for users to express their opinions or share pieces of interesting news by posting relatively short messages (corpus) compared with the regular blogs. The volume of corpus updates that users receive daily is overwhelming. Also, as information diffuses from one user to another, some topics become of interest to only small groups of users, thus do not become widely adopted, and could fade away quickly. This paper proposes a framework to enhance user's interaction and experience in social networks. It first introduces a model that provides better subscription to the user through a dynamic personalized recommendation system that provides the user with the most important tweets. This paper also presents TrendFusion, an innovative model used to enhance the suggestions provided by the social media to the users. It analyzes, predicts the localized diffusion of trends in social networks, and recommends the most interesting trends to the user. Our performance evaluation demonstrates the effectiveness of the proposed recommendation system and shows that it improves the precision and recall of identifying important tweets by up to 36% and 80%, respectively. Results also show that TrendFusion accurately predicts places in which a trend will appear, with 98% recall and 80% precision.

Twitter has emerged as one of the most powerful micro-blogging services for real-time sharing of information on the web. A large base of Twitter users tend to post short messages of 140 characters (Tweets) reflecting a variety of topics. Location-based-services (LBSs) may be built on top of microblogs to provide for targeted advertisement, news recommendation, or even microblogs personalization. Knowing the user's home location would empower such LBSs. In this paper, we propose prediction models to infer the users' home location based on their social graph and tweets content. The problem is non trivial as the tweets are short and not many people like to share their location for privacy concerns. Our extensive performance evaluation on a publicly available dataset demonstrates the effectiveness of the proposed models. The proposed models outperform the competitive state-of-the-art home location inference techniques that are based on the social graph, tweet content, and both by a relative gain in the F-measure of up to 37.71%, 29%, and 9.06%, respectively.

Abstract Scientific data acquired through sensors which monitor natural phenomena, as well as simulation data that imitate time‐identified events, have fueled the need for interactive techniques to successfully analyze and understand trends and patterns across space and time. We present a novel interactive visualization technique that fuses ground truth measurements with simulation results in real‐time to support the continuous tracking and analysis of spatiotemporal patterns. We start by constructing a reference model which densely represents the expected temporal behavior, and then use GPU parallelism to advect measurements on the model and track their location at any given point in time. Our results show that users can interactively fill the spatio‐temporal gaps in real world observations, and generate animations that accurately describe physical phenomena.

Context is any information used to characterize the situation of an entity. Examples of contexts include time, location, identity, and activity of a user. This paper proposes a general context-aware DBMS, named Chameleon, that will eliminate the need for having specialized database engines, e.g., spatial DBMS, temporal DBMS, and Hippocratic DBMS, since space, time, and identity can be treated as contexts in the general context-aware DBMS. In Chameleon, we can combine multiple contexts into more complex ones using the proposed context composition, e.g., a Hippocratic DBMS that also provides spatio-temporal and location contextual services. As a proof of concept, we construct two case studies using the same context-aware DBMS platform within Chameleon. One treats identity as a context to realize a privacy-aware (Hippocratic) database server, while the other treats space as a context to realize a spatial database server using the same proposed constructs and interfaces of Chameleon.

In order to enable extensibility, modern query optimizers typically leverage a transformation rule based framework. Testing individual rule correctness as well as correctness of rule interactions is crucial in verifying the functionality of a query optimizer. While there has been a lot of work on how to architect optimizers for extensibility using a rule based framework, there has been relatively little work on how to test such optimizers. In this paper we present a framework for testing query transformation rules which enables: (a) efficient generation of queries that exercise a particular transformation rule or a set of rules and (b) efficient execution of corresponding test suites for correctness testing.

In this paper, we introduce PLACE*, a distributed spatio-temporal data stream management system for moving objects. PLACE* supports continuous spatio-temporal queries that hop among a network of regional servers. To minimize the execution cost, a new Query-Track- Participate (QTP) query processing model is proposed inside PLACE*. In the QTP model, a query is continuously answered by a querying server, a tracking server, and a set of participating servers. In this paper, we focus on query plan generation, execution and update algorithms for continuous range queries in PLACE* using QTP. An extensive experimental study demonstrates the effectiveness of the proposed algorithms in PLACE*.

Rank-aware query processing has emerged as a key requirement in modern applications. In these applications, efficient and adaptive evaluation of top-kqueries is an integral part of the application semantics. In this article, we introduce a rank-aware query optimization framework that fully integrates rank-join operators into relational query engines. The framework is based on extending the System R dynamic programming algorithm in both enumeration and pruning. We define ranking as an interesting physical property that triggers the generation of rank-aware query plans. Unlike traditional join operators, optimizing for rank-join operators depends on estimating the input cardinality of these operators. We introduce a probabilistic model for estimating the input cardinality, and hence the cost of a rank-join operator. To our knowledge, this is the first effort in estimating the needed input size for optimal rank aggregation algorithms. Costing ranking plans is key to the full integration of rank-join operators in real-world query processing engines.Since optimal execution strategies picked by static query optimizers lose their optimality due to estimation errors and unexpected changes in the computing environment, we introduce several adaptive execution strategies for top-kqueries that respond to these unexpected changes and costing errors. Our reactive reoptimization techniques change the execution plan at runtime to significantly enhance the performance of running queries. Since top-kquery plans are usually pipelined and maintain a complex ranking state, altering the execution strategy of a running ranking query is an important and challenging task.We conduct an extensive experimental study to evaluate the performance of the proposed framework. The experimental results are twofold: (1) we show the effectiveness of our cost-based approach of integrating ranking plans in dynamic programming cost-based optimizers; and (2) we show a significant speedup (up to 300%) when using our adaptive execution of ranking plans over the state-of-the-art mid-query reoptimization strategies.