All the abstracts and full papers submitted to iCATIS2019 has been assessed by an independent review panel and published in the Conference Proceedings with an ISBN Number 978-0-6482681-6-1.

Abstract: The term computer industry refers to business that involves computing technology - design of computer hardware and computer networking infrastructures, development of computer software, manufacture of computer components, providing computer related technical consultancy, services and providing information technology services (ITS). Broadly, Computer Industry is referred as Information Technology (IT) industry as it not only encompasses computing technologies but also communication technologies that involve exchange of information.
The rapid expansion of the IT industry in India has been a key feature in the country’s economic development. Information Technology in India is an industry consisting of two major components: IT services and business process outsourcing (BPO). The sector has increased its contribution to India's GDP from 1.2% in 1998 to 7.7% in 2017. According to NASSCOM, the sector aggregated revenues of US$160 billion in 2017, with export revenue standing at US$99 billion and domestic revenue at US$48 billion, growing by over 13%. India’s IT industry contributed around 7.7 per cent to the country’s GDP. IT industry is fuelling the growth of startups in India, with the presence of more than 4,750 startups in India. View

Abstract: Emerging technologies such as mobile computing, sensors and sensor networks, and augmented reality have led to innovations in the field of wearable computing. Devices such as smart watches and smart glasses allow users to interact with devices worn under, with top of clothing and a new paradigm of computing is emerged. The many opportunities offered by wearable computing have triggered the imaginations of designers and researchers in a wide variety of fields. The inevitability of computers and interfaces which are small enough to be worn on the human body has inspired the creation of devices and applications which can assist with specialized professional and personal activities. Wearable computing considers the use of miniature computing devices in order to support various human activities. Such devices can be utilized similarly to a more established application of mobile learning in order to enable learning independently of temporal and spatial constrains. Although some learning management systems can support mobile learning through the use of adaptive themes or custom learning activities, wearable computing introduces many new variables into the mix, and learning management systems are not equipped to support such devices. In this talk, a detail overview of wearable technologies will be presented from the perspectives of future innovation and information security. Download

Abstract: Jamming attack is a serious security threat for Wireless Sensor Network (WSN) in which adversaries intentionally emit radio frequency signals to corrupt ongoing transmission. Several anti-jamming techniques have been devised as countermeasures against jamming attacks. These techniques detect jamming relying on the parameters such as Packet Delivery Ratio (PDR), Packet Send Ratio (PSR), Received Signal Strength Indication (RSSI) and Clear Channel Assessment (CCA). On the other hand, energy efficiency is the major issue for WSN and it is a dominating factor for network lifetime. To prolong network lifetime, WSN is organized into clusters where each cluster head transmits aggregated data to the base station directly or using multi-hop. This paper focuses on jamming attack at the network layer in a clustering based WSN. Jamming attack in a clustered WSN affects drastically and makes clusters ineffective. As a result, WSN may experience delay in transmission, and reduction in network throughput. To overcome jamming attacks, we propose an effective jamming localizing scheme where parameters such as Jamming Signal Strength (JSS), PDR, etc., are considered in detecting jamming attacks. To continue transmission in presence of jamming, we also propose a cluster-based channel assignment technique to avoid the channels that are thwarting transmission efficiency. The simulation results reveal the accuracy of our proposed scheme in identifying jamming region, reduction in packet loss due to jammer's presence, and better throughput. [pp. 4-10] Download

Abstract: Big data is now a widely talked issue. Big data is defined by the information assets characterized by such a high volume, velocity and variety to require specific technology and analytical methods for its transformation into value. Some classical examples of Big Data are call logs, mobile-banking transactions, online user-generated content such as blog posts and Tweets, online searches, satellite images, etc. These actionable information requires using computational techniques to unveil trends and patterns within and between these extremely large socioeconomic datasets.
The Sustainable Development Goals (SDGs) were adopted by the UN general Assembly in September,2015 with 17 goals 169 targets and 232 indicators to be achieved between 2016-2030.In order to monitor the progress of SDG indicators a wide ranges of data are required. The traditional censuses and surveys are not sufficient to cater the data need of monitoring SDG. Therefore, the United Nations Statistical Commission (UNSC), who are responsible for preparing indicator frame work, methodology and data source, recognized Big data as a source of information. So, Big data can be used for monitoring selected SDG indicators in Bangladesh.
The National Statistical Organization (NSO) like BBS have to adopt new technologies and methods to enable sound statistical analysis of Big data so that selected SDG indicators can be obtained from big data which will be cost –effective and disaggregated at sub national levels.
This paper will highlight the possibility of such measures.[pp. 11-20] Download

Abstract: Entity Resolution (ER) is a prerequisite to several Web applications including enhancing semantic searches and information extraction from the Web, strengthening the Web of Data by interlinking entity descriptions from autonomous sources, and supporting reasoning using related ontologies. While designing an ER system, it is assumed that each entity profile contains of a uniquely identified set of attribute-value pairs, each entity profile matches to a solitary real-world object, and two similar profiles are identified as long as they co-occur in at least one block. ER is an inherently quadratic problem (i.e., O (n2)), given that every entity must draw a comparison with others. ER does not scale to large entity collections, as in Web data. The most well-known solution for addressing large-scale ER in the literature is blocking, which is an approximate solution where similar entities are grouped into blocks and comparisons are limited to within blocks. The process of entity resolution and the types of entity resolution in relational and Web data are discussed in this paper. Further, the paper reviews the literature on the approaches introduced by former researchers on the entity resolution system. The data integration, block building, and block processing phases, and the challenges involved for designing an efficient ER system are discussed. This paper concludes with the measures required to evaluate entity resolution approaches. Download

Abstract: An Automated Fever Detection System (AFDS) has been proposed in the paper that will extract symptoms from patient’s physical problem description in natural language. Symptoms extracted shall be able to detect the fever type in its early stage. Symptoms and synonyms databases have been associated with this AFDS. AFDS uses NLP techniques to detect the symptoms from unstructured text which means symptoms are extracted from physical problem description in natural language which is given as an input. The input text undergoes tokenization, POS tagging, substring formation, combination creation, repetition removal and then symptom extraction. The extracted symptoms are further matched with the synonyms database in the proposed architecture and the complete execution of the algorithm detects the disease of the patient. The Fever Detection model can be used in detection and diagnosis of diseases from the symptoms of fever. It may act as a doctor’s assistant. It may be modified and extended to address several complicated diseases. [pp. 29-35] Download

Abstract: Human action recognition has received a significant amount of attention, especially after the breakthrough of two-stream-based action recognition proposal. In the two-stream-based action recognition, it used spatial and temporal networks and predicted an action. The spatial network used features from sequential image frames of video data. On the other hand, the temporal network used features from the optical flow of video data. However, we have found that human postures can also be good features for action recognition. In our research, we incorporated posture-based image features and trained a further network for action recognition. Experiments show that our proposal outperforms state-of-the-art frameworks. Download

Abstract: The process of identifying a part-of-speech (POS) for a word is called Part-of-speech tagging. POS tagging is an essential task in natural language processing. For example, speech recognition, sentence parsing, information retrieval, and information extraction etc., use POS tagging. To identify part-of-speech, researchers use different POS tagging techniques like rule-based POS tagging, statistical class model-based POS tagging etc. Usually, these systems make use of different contextual information of words along with the variety of word-level features for identifying POS tag. However, for languages with rich morphology, generic part-of-speech tagging algorithms do not yield very high accuracy. The reasons are the data sparseness issue and the grammatical dependencies of words hidden in the sentence, which are very essential to identify a POS tag. In our research, we propose to develop a language-independent deep part-of-speech tagger that can identify part-of-speech by considering grammatical dependencies and features of words hidden in the sentence. These grammatical dependencies and features are fed into a neural network-based classification model. We consider only 17 part-of-speech tags that are available in most of the languages. Our language independent deep part-of-speech tagger performs better than the state-of-the-art part-of-speech taggers of different languages. Download

Abstract: Machine learning is a branch of Artificial Intelligence. Machine learning techniques have achieved a huge success in a wide area of applications, such as image classification, speech recognition, text classification. Machine learning plays a big role in our everyday life by using numerous applications. The cores of these applications are designed with a number of machine learning algorithms. In this paper, we review a number of algorithms which are very popular in the past and recent times. Download

Abstract: Rapid Growth of population and vehicles in cities forges smart parking management for vehicles imperative. Smart parking system has become prerequisite for turning the cities into smart cities, especially in this era of IoT (Internet of Things). Besides that, loads of data perceived by the sensors of IoT environment makes it harder to suggest the best solutions to the user for his problem. An architecture has been proposed in this paper where the best parking solution will be suggested to the user using machine learning approach. A Naïve Bayes classifier will classify user data based on a dataset produced from the data sensed by the sensors. A single board computer (Raspberry Pi) will work as IoT core device and will send the sensed data to the server rapidly. The server will generate a dataset with these data and will learn through this dataset. A web app will be provided to the user where he will be requested to give some information like parking time, preferred space, parking fee etc. The Naïve Bayes classifier then perform the classification of user data in the server side using the training dataset and will suggest the best parking place. Download

Abstract: Car accidents are considered one of the most destructive phenomena. Though there are many different reasons behind car accidents, most accidents occur due to driver’s unawareness and uncontrolled speed. Also, there seems to be a problem reaching the spot of accident in time for lack of awareness. As a solution, the advent of Internet of Things (IoT) technologies can reduce the number of accidents. In this paper, a smart system is described that alerts and controls the speed of a vehicle, also notifies the individuals accordingly when an accident occurs. This system always monitors the distance between vehicles and obstacles that are in front, using distance sensor. It will alert the driver to control the speed and reduce the speed by itself when a critical distance comes. Whenever an accident takes place for uncertain condition, an email alert will be sent to the accountable individual with car details. Download

Abstract: Garbage management is a big challenge in today’s world. Respective authority has to pay a lot of money for the manual garbage collection and management every year but the situation remain same. We often see the scenario of overflowed garbage bins everywhere which causes bad impact on the environment and increases the health risks of nearby citizens. Here we represent a better solution with the help of IOT and given first priority to our environment. For this we use RFID reader for tracking the waste dumper as well as the user authentication which interact with a web-based online system. On the basis of added waste weight, our system calculate a reward for the particular user and update it into the database. It also measures the level of the dustbins and update it into the main server with a regular interval. When the level of dustbins overcome the actual threshold value the system automatically select those bins and creates an optimized route. And then, according to the capacity of the waste collecting vehicles, the system notifies the bin status to the waste collectors with the location of all bins from the map. And they can check all status through an android application. Furthermore, our proposed system can give prediction result on the basis of different parameter which helps the authority for future decision making. From this work we set our goal is to get a better solution beyond the existing system, consuming total waste management cost, save the environment, encourage people, select high polluted area and collecting waste in a smart way. Download

Please click here to down the proceedings.