Abstract In software, an algorithm is a well-organized sequence of actions that provides the optimal way to complete a task. Algorithmic thinking is also essential to break-down a problem and conceptualize solutions in some steps. The proper selection of an algorithm is pivotal to improve computational performance and software productivity as well as to programming learning. That is, determining a suitable algorithm from a given code is widely relevant in software engineering and programming education. However, both humans and machines find it difficult to identify algorithms from code without any meta-information. This study aims to propose a program code classification model that uses a convolutional neural network (CNN) to classify codes based on the algorithm. First, program codes are transformed into a sequence of structural features (SFs). Second, SFs are transformed into a one-hot binary matrix using several procedures. Third, different structures and hyperparameters of the CNN model are fine-tuned to identify the best model for the code classification task. To do so, 61,614 real-world program codes of different types of algorithms collected from an online judge system are used to train, validate, and evaluate the model. Finally, the experimental results show that the proposed model can identify algorithms and classify program codes with a high percentage of accuracy. The average precision, recall, and F-measure scores of the best CNN model are 95.65%, 95.85%, and 95.70%, respectively, indicating that it outperforms other baseline models.

Despite the incredible adoption of cryptocurrencies, blockchain-based cryptocurrencies have likewise raised some concerns. The scalability problem is the major one among them. An off-blockchain payment channel network (PCN) has been introduced to solve this issue. PCN can fundamentally reduce blockchain scalability by constructing a number of payment channels between the nodes and without committing every single transaction to the blockchain. But as a matter of fact, there has an unwanted assumption in PCN that channel participants must remain online and follow blockchain updates, for the synchronization with blockchain to protect the channel against deception. To mitigate this issue “Watchtower” concept has been proposed. Watchtower is a watching service and always stays online that a channel participant can hire it by offering incentives for monitoring the channel and checking blockchain updates consistently to prevent fraud on behalf of the hiring party. However, watchtower may be more beneficial by cooperating with the cheating counterparty and neglecting to perform the watching service properly. The efficiency drawback can occur for that. In this work, we have been motivated by this issue and tried to find out an effective and reliable watchtower for the channel watching service from multiple watchtower nodes or candidates in the PCN. In particular, we have been approached by using the distributed Peterson Leader-Election Algorithm to find the best watchtower among multiple of them where the more successfully performed work node or candidate will be selected for the channel monitoring job. We also have provided a detailed step-by-step process of the algorithm including experiments and illustrations for employing watchtower among multiple of them.

The development and operation of Online Judge System (OJS), which is used to evaluate the correctness of programs, is a nontrivial and difficult task due to the various functional and non-functional requirements. However, although many OJSs have been developed and operated, and their usefulness reported, the theory for constructing OJSs has not been sufficiently discussed. In this paper, we present the functional and nonfunctional requirements oriented to OJS as well as demonstrate the internal components and software architecture of an OJS, which has been in operation for over a decade and has evaluated over six million solutions. We also present real-world experiences and challenges encountered during this long journey of our OJS.

In recent years, vast amounts of unstructured data have been and are being produced from multiple sources around the world. In the field of data mining, clustering is the most efficient technique for grouping such unstructured or unsupervised data. In its basic form, data clustering is an unsupervised method that groups or cluster objects so that all objects within the same cluster are very similar to each other, whereas objects grouped similarly in the different clusters are quite distinct. However, due to the exponential growth of data amounts available in a wide variety of scientific fields, it has become increasingly difficult to manipulate and analyze such information. In addition, it is becoming progressively more cumbersome to extract hidden features from the resulting huge unstructured and unsupervised datasets. This study reports on an improved general k-means clustering algorithm that was created by modifying its initial centroid selection process and adding a new outlier detection and filtering algorithm. Under normal conditions, most algorithms select initial centroids randomly, which often leads to poor initial cluster quality. Additionally, most existing outlier detection techniques are inadequate due to their poor accuracy levels, high computational complexity, and inability to identify outlier data. In contrast, after an analysis of comprehensive experiments performed to validate our approach via comparisons against existing techniques and benchmark performance values, we found that our proposed approach performs better than existing methods in terms of initial centroid selection, outlier detection, and other related matters.