\documentclass{sig-alternate-05-2015}
\usepackage{etoolbox}
\makeatletter
\patchcmd{\maketitle}{\@copyrightspace}{}{}{}
\makeatother

\begin{document}

\title{Web Science \& Engineering - Homework(Q2)}
% Copyright
\setcopyright{acmcopyright}


\author{
\alignauthor
Paras Kumar\\
       \affaddr{Delft University of technology}\\
       \email{p.kumar-6@student.tudelft.nl}
}
\maketitle
\section{Homework 3}
\paragraph{\textbf{3.1 Explain what is a User Model. (min. 100 words)}}
User model is defined as the collection and classification of personal data associated with a particular user. It’s a data structure that is used to store certain information about the behaviour of an individual user based on the actions and activities performed within the context of our product and user profile is the real representation in a given user model. Briefly, user model is the technique for achieving and illuminating user profile. The main goal of the model is sketching the customization and adaption of the system to the user’s specific needs. With the help of this user model mechanism, it is trivial to  create a more engaging experience for each user individually.
\paragraph{\textbf{3.2 Explain what are User Adaptive systems. (min. 100 words)}}
In web science, Adaptive systems refers to a process in which an interactive system adjust its behaviour to individual users based on information acquired, context of use and surroundings. Based on the information about user model to adapt a system or interface in such a way that it enriches the system functionality and experience of the end user and keep learning with further data. There are many examples of such adaptive systems which are extensively used by the users. For example the news feed of facebook, twitter, instagram, dark/light mode of smartphone depending on the sunlight, personalization of items on the e-commerce store, sensors of temperatures etc. All these adaptive systems enable users in decision making and help them in discovering similar items which they never search or thought before.
\paragraph{\textbf{3.3a Give an example of how user modeling can be practically done and applied in the context of online education. (min. 200 words)}}
Web based online education is a hot research and development area. Online education is class independent and it’s easy to learn about each individual user profile and deliver content accordingly based on the different aspects. For example level of the user, goals of the user, the speed of capturing lecture and feedback  based on the questions asked during the interactive lecture in the form of quiz, assignment, opinions etc. Many popular online websites such as coursera, udacity, edx and etc are delivering online education and based on the user profile recommending courses which perfectly fits individual career goals. In the past it was not possible to have online education as the internet makes it accessible to every user connected with internet. In the online learning classroom we can keep track of individual user behaviour and based on it, it’s easy to recommend and suggest useful suggestion on his performance. Feedback and reviews from students are true very important in improving such systems, the more we know about our end users the much better content we can deliver. In modern tutorials several poll questions are asked during video lecture in order to learn about each and every aspect of the lecture.
\paragraph{\textbf{3.3b For the example from 3.3a, describe what scientific research could contribute (= what would scientific research do for that example?) (min. 200 words)}}
Online educational websites are modernized these days as compared to several years ago. With the development of new scientific methods and algorithms in the field of data science, machine learning, big data the learning experience for understanding end user is keep evolving. With the huge amount of data collected from reviews, feedback; recent data science algorithms can be applied to discover hidden patterns in our data. The adaption of systems based on users behaviour is also determined in an efficient manner with the power of artificial intelligence. User interface design and experience techniques are also improved in online education with the help of user interaction and engagement on our course content. On the contrary, online courses provide access to students who would never have the opportunity or inclination to take classes in person. The community of scientists behind these cutting edge technologies help the end users to get better and improved online content experience day by day. From hardware to software each component is involved in online education ecosystem. The connectivity of devices and IOT technology will improve further the web based educational system. The real time analysis of feedback and reviews will enable users to get right answers in short span of time.
\paragraph{\textbf{3.3c Explain how Web Data can be as a source in the scientific research for 3.3b. (min. 150 words)}}
In the modern age of technology data is the new fuel for everything in the daily life actions. Web data collected via distance learning platforms will help us to understand the needs of the individual user’s by applying semantic language processing techniques. We can train our models of machine learning and deep learning to predict the results for optimizing behaviours among these platforms. The way of delivering knowledge is very critical for users because it’s always hard to hold the engaging power of the user in such a monotonic environment, hence the formation of story and graphical representations will be more powerful technique to make it user oriented globally. Based on the geographical locations and culture we can learn different tips and tricks for delivering better content across the world in their own local language based on the subtitle obtained via language translation API’s available. Web data via reviews and feedback is useful for predicting and enhancing the user experience in online education.
\paragraph{\textbf{3.3d Read one scientific paper that is concerned with scientific research on user modeling (learner modeling) in online education, e.g. in MOOCs, and describe its main research questions and contributions. (min. 200 words)}}
The paper is \cite{cook2015mooclm}
\\
The idea of one to one personalized e-learning experience is emerged from the recommendations and suggestions on the e-commerce websites. For a single teacher in real class room it’s hard to focus on individual and at the same time class has diversity of minds. Some students are way advanced while others are struggling with basic content. This paper proposed a six layered framework for applying personalized e-learning recommendations based on the level, background and attitude of the student. The six step framework is summarized as follows. 1st step is learning behaviour of user with the help of questionnaire. 2nd step includes the formatting of data. 3rd step is important and it implements the adaption module by choosing available recommending techniques. 4th step involves learner model which focuses on the learning style of students and based on the student it decides the available learner model, which fits the situation perfect. 5th step is the domain model and 6th step is generating actual recommendation items to the user. The personalized e-learning systems takes into consideration the learner style and personality for learning material recommendation per each student. The suggested framework solves the previous literature limitations like cold start problem, sparsity problem, and the over specialization problem.
\paragraph{\textbf{3.4a Give an example of how user modeling can be practically done and applied in the context of urban analytics and urban (big city) challenges. (min. 200 words)}}
There are numerous services offered by the government offices in the big cities and it’s important to benefit the inhabitants of the big cities as they are contributing main role in the economy of the region. A small use case of implementing rating mechanism of the services offered based on the devices in the offices. It will enable government organizations to analyze the personality of the citizens and their satisfaction level within the country. Based on the problems reported by locals in the context of health, transport, waste management, crime, education etc. the government can take actions in real time and show performance to the citizens by pushing updates to their devices. It will create a transparency between users and organizations and will empower them to be sincere with their country and encourage tourism towards their cities globally. Usually these types of actions take a lot of time and it’s time consuming process to follow up on the progress of the individual tasks. With the devices installed everywhere in public places along with organizations government can learn about their behaviour and take action in real time based on the report. Handling bias data will be a challenging task in this case but it can be solved by introducing identity and polling mechanism.
\paragraph{\textbf{3.4b For the example from 3.4a, describe what scientific research could contribute (= what would scientific research do for that example?) (min. 200 words)}}
The scientific methods played a key role in refactoring the technology the way it looks today. Smart cities are the birth of technology and the whole infrastructure needs hardware and software components to fulfil the needs of the citizens in an efficient manner. Various activities of crime detection by cameras, traffic signal violations are detected by cameras with the help of powerful algorithms in the domain of computer vision and  data science models. The learning models can be estimated accurately with the huge amount of data generated by citizens. In order to process and store this huge amount of data big data frameworks techniques can be applied. In order to reduce the latency of processing and generating results in real time IOT and 5g will put this in the right direction. All this analytics of data help organizations and businesses to learn about their potential users. Understanding the behaviour of the citizens is crucial for taking steps in the further development of the cities and by leveraging the power of technology and scientific approaches we can build the cities better and revamp life of the citizens. Challenges are involved to cater all these problems in real life but it will ensure high rate of transparency.
\paragraph{\textbf{3.4c Explain how Web Data can be as a source in the scientific research for 3.4b. (min. 150 words)}}
All the revolutions in the technology comes with the power of data and it’s accessible to everyone via the web. It’s a decentralized store of information which is not controlled by any single individual and everyone can contribute to that. The web data will help the organizations in mining the blog posts, tweets, feedback and reviews from the web in order to learn about the problems of the city and help their citizens to get rid of those hassels. All the results are generated based on the web data available and accessible to the government organizations publicly and privately. Processing of web data has many challenges such as processing of different formats of data, storage and real time streaming of data in critical situations to detect some results. Machine learning models and neural networks are trained with the help of data generated by the users and it’s collected from web.
\paragraph{\textbf{3.4d Read one scientific paper that is concerned with scientific research on user modeling in urban analytics/challenges, e.g. around mobility, and describe its main research questions and contributions. (min. 200 words)}}
The paper contributes an algorithm which assists elder people while navigating through environment using robotic walker instrument. The walker develops the hierarchical Markov model operates at three different levels by learning daily walking activities. First level is metric motion level at which location is described by metric coordinates. Second a topological motion level which uses topological regions as its basic element and last is activity level, at which a person’s walking activities are logically subdivided into broader categories. Using supervised classification method walker model is trained. Two different proto-types are proposed and both are equipped with a laser-based navigation system, a display for providing directions to its users, a touch-based interface for receiving commands, and an active drive mechanism equipped with a clutch for switching between active and passive mode. The guidance provided by the walker is resembling to the car-based GPS systems. It is important to use the power of technology for solving the problem of people who need assistance. The mobility of old age people becomes independent by using this system. In the future, the plan is to utilize the learned models in our guidance system, in the hope of providing the right guidance at the right time even if a person fails to specify the target location.

\maketitle
\section{Homework 4}
\paragraph{\textbf{4.1 Consider one running example (you choose) of a Web-based information system that incorporates concepts from the Social Web. Imagine you are responsible for (design and use of) that system.}}
The proposed example will lie in the domain of “New Form of Communities”. One of the real problems I faced being an international student is how to interact with other peers in the class coming from different backgrounds. I spend my 1st year of studies in Berlin and barely made any german friend and it’s been four months in the Delft and I didn’t get a chance to meet any dutch student. Apart from this, there are many international students I can meet but how can I discover the interested students in the class who is willing to interact. A social media platform can be developed for solving this problem and it will help us to find friends from different backgrounds and cultures.
\paragraph{\textbf{a. Describe that system, as sufficient context for the next questions.}}
The idea is to develop an application and website which can help in connecting with peers in the university. Students will be able to create their profile and expressing interests, expertise and background. Instead of sharing stories and timeline posts students will share their local experience with their fellow students. It will help them to build a strong community and discovering more fun together. It will also be useful in formulating their groups in course projects as they already know each other better. This differs from facebook, instagram and snapchat because on these social media platforms people prefer to connect with their close friends and community people they already know but this application will enable them to connect with strangers without worrying about interrupting their existing timelines on other platforms.
\paragraph{\textbf{b. For each of the five concepts mentioned on slides 28-32 (new means of communication, new forms of communities, new source of knowledge, a new source of entertainment, new venues for self-expression) describe how they could look like in your running example.}}
The proposed solution will extend the existing power of technology for communication and enabling connections between students as a new source of medium connecting with people. Once students made friends using this platform they can form communities among each other and have fun together, made plans for helping societies and share stories with each other. It improves end user knowledge by getting familiar with each other's cultures and expanding their network. The student life is also about organizing local events and parties together, this platform can help them in building teams and making it happen, also user can post their funny cultural games and stories to attract their peers. In a nutshell, this is similar to other existing social media platforms but the way of connecting them is different as it will only target the students within the university. The user verification can be performed by uploading university cards. Junior students can discover seniors who already take specific courses and learn from their experiences and have a change to get the guidance. Using this platform we can connect many people together and made stronger community for the betterment of this society.
\paragraph{\textbf{c. For each of the eight concepts mentioned on slides 39-46 (personal profiles, establishing connections, participating in groups, communicating with connections, sharing content, expressing opinions, finding information, retaining users) describe how they could look like in your running example.}}
In the suggested application, users can create their personal profile and based on their profile attributes other students will appear in their “Discover Peers” section. The profile of the user is not private as everyone is stranger on this platform and curious to know about each other in order to find students of similar interests around the campus. As there are various goals of the students such as studying together, hangout for fun, strolling around the city etc. In the extension, we can form a group based on the department in the universities which help them to discover useful events, activities and interesting projects happened or happening in the department. The communication in this channel is critical when you don’t know each other hence based on the premium subscription students will send a direct message to limited number of students on each day. Content will not be shared explicitly as a time but it will be inside the user timeline which anyone can visit explicitly but only friends will be able to like and comment. The idea of timeline is a bit different from facebook and instagram because in this context no one is interested in viewing the timelines of strangers but once you have an interest in the profile then you can visit his/her timeline. With the help of timelines students can discover content and express their interesting stories of campus life before and ongoing. Every year new students are coming to the university so it will be easy to obtain the users every year. In order to hold the retention power of users this application can be extended to create an alumni section for each university, with which they can discover their friends in the industry and relevant companies. Again the target will be students community and more importantly they can also discover interesting events in other universities around the world on one platform in a centralized manner.
\paragraph{\textbf{4.2 Read one scientific paper that is concerned with scientific research on crowdsourcing or human computing and describe its main research questions and contributions. (min. 200 words)}}
The paper “crowdsourcing images for global diversity” aimed to create a dataset of images captured from around the world of distinct locations. An existing microtasking android mobile application was used to collect real world labelled images via micro crowdsourcing and this dataset is used for model training. Users can flexibly choose between tasks and to engage users in task completion, the system leverages game mechanics suggested by prior work. In our design, logged-in users receive badges for answering questions of different categories and get to level up by contributing more answers. A micro crowdsourcing platform is developed where users can open source their work. Based on the result, it is observed that users are motivated to organize social events and capture the moments. In future, this kind of crowdsourcing efforts can help us to document more events and share the knowledge around the world. Users are also interested in getting feedback on their work to know that “How well they are contributing to this community and it pushes them to do more”. The developed image classifier model will be useful to predict the class of image in a broader context. There are several challenges involved in detecting bias in crowd sourced data, which is not part of this paper.
\maketitle
\section{Homework 5}
\paragraph{\textbf{5.1 Consider one example of a scientific study of a societal phenomenon where data from the Social Web is used: “data science with social web data”.
}}
Data scientists are learning constantly to tell surprisingly accurate stories about people from that accumulation of information. The analytics of social web data enables businesses and marketing researchers to understand the behaviour of user interaction with various posts, videos, images and etc. In today’s world all these social media platforms  are creating vast amounts of data and it is considered as a core valuable asset for the company. Meaningful insights of users are determined from social web data. So Facebook, Twitter, Instagram, and other prominent social media sites dove into Big Data almost immediately for reasons identical to those of any other retailer: the better they understand their audience, the more appealing they can make their product.
\paragraph{\textbf{a. Describe the [driving] research question.
}}
Social media networks allow you to reach everyone and anyone, wherever they are in the world. That’s a good thing if you want to have a lot of followers. But as a brand you are not necessarily interested in getting likes from Sydney-based Katie or Vincent from France. They may be great people, but hardly relevant prospects for a local travel agency in Leeds. Social networks present us with a new challenge - “how do I find my ‘tribe’ of customers and connect with them based on demographics”
\paragraph{\textbf{b. Describe the intuition for why [a specific way of using] social web data can provide meaningful insights on that societal phenomenon [fitting with the research question].
}}
Social media platforms allow users to express their opinions and experiences. It’s a quicker way to learn about the reviews of target audience and most accessible way to reach as many people as we can. These days many companies are running digital marketing campaigns. Using twitter public API’s, we can extract the target profile of users of a particular region and based on their mindset, social aspects and nature we can tweak our product accordingly. There are many attributes of audience which are critical in determining the key features of the product such as language, gender, age etc. One example is creating a social media platform proposed in previous homework is we can only target the cities which has universities, if something is launched with respect to the tourism then we can target most popular tourist spots of the world visited by travelers and the trend can be determined by social media posts and hashtags.
\paragraph{\textbf{c. Describe one thing that could go wrong in this case of data science.
}}
The quality of online data being analysed is always a concern among enterprises. Social media platforms are littered with fake and duplicate profiles. Adding to the problem are the access restrictions on most profiles that make it difficult to verify their validity. Also, social channels provide very little or inaccurate information on user journey. The handling of bias data and fake news is a major challenge in model training in data science. If we have a lot of bias in data then decisions can go wrong. Researchers are trying to predict these things via crowdsourcing but again the quality and variety of data is a major concern.
\paragraph{\textbf{d. Read one scientific paper related to this example subject, give its reference, and summarise in about 10 sentences what the key message of this paper is.}}
This paper proposed an approach for contextual big analytics in social web data, particularly in twitter as it is one example of a real time data source. The combination of rich context model with machine learning is used in order to improve the quality of data mining techniques. The analytical results will help us in context aware recommendations. With the suggested example we can target specific group of users using this approach. Thus, context information is very important to consider in data analysis in order to provide better decision-making, predictions and recommendations. The context of tweet is extracted from metadata and integrating into a context model in a machine-learning algorithm in order to cluster tweets by their contextual similarity. Apache Spark is used as an underlying framework for the real time processing of data. Extracting context information from streaming data and processing it in real-time, offers new possibilities for big data analytics. Thus, we believe that adding context information about the tweet may bring new opportunities for designing advanced analytics services. Later on, it might be possible to analyse these clusters of tweets in context and discover some additional contextual parameters useful for further analysis.

\bibliographystyle{abbrv}
\bibliography{sigproc}
\end{document}