Matching Items (4)
Filtering by
- All Subjects: Amazon Web Services
- All Subjects: Spotify
- Creators: Balasooriya, Janaka
- Creators: DiMuro, Louis
- Creators: Koudssi, Zakaria Corley
- Status: Published
Description
A common challenge faced by students is that they often have questions about course material that they cannot ask during lecture time. There are many ways for students to have these questions answered, such as office hours and online discussion boards. However, office hours may be at inconvenient times or locations, and online discussion boards are difficult to navigate and may be inactive. The purpose of this project was to create an Alexa skill that allows users to ask their Alexa-equipped device a question concerning their course material and to receive an answer retrieved from discussion board data. User questions are mapped to discussion board posts by use of the cosine similarity algorithm. In this algorithm, posts from the discussion board and the user’s question are converted into mathematical vectors, with each term in the vector corresponding to a word. The values of these terms are computed based on the word’s frequency within the vector’s corresponding document, the frequency of that word within all the documents, and the length of the document. After the question and candidate posts are converted into vectors, the algorithm determines the post most similar to the user’s question by computing the angle between the vectors. With the most similar discussion board post determined, the user receives the replies to the post, if any, as their answer. Users are able to indicate to their Alexa device whether they were satisfied by the answer, and if they were unsatisfied then they are given the opportunity to either rephrase their question or to have the question sent to a database of unanswered questions. The professor can view and answer the questions in this database on a website hosted by use of Amazon’s Simple Storage Service. The Alexa skill does well at answering questions that have already been asked in the discussion board. However, the skill depends heavily on the user’s word choice. Two questions that are semantically identical but different in phrasing are often given different answers. This is because the cosine algorithm measures similarity on the basis of word overlap, not semantic meaning, and thus the application never truly “understands” what type of answer the user desires. Improving the performance of this Alexa skill will require a more advanced question answering algorithm, but the limitations of Amazon Web Services as a development platform make implementing such an algorithm difficult. Nevertheless, this project has created the basis of a question answering Alexa skill by demonstrating a feasible way that the resources offered by Amazon can be utilized in order to build such an application.
ContributorsBaker, Matthew Elias (Author) / Chen, Yinong (Thesis director) / Balasooriya, Janaka (Committee member) / Historical, Philosophical & Religious Studies (Contributor) / Computer Science and Engineering Program (Contributor) / Barrett, The Honors College (Contributor)
Created2019-05
Description
Music streaming services have affected the music industry from both a financial and legal standpoint. Their current business model affects stakeholders such as artists, users, and investors. These services have been scrutinized recently for their imperfect royalty distribution model. Covid-19 has made these discussions even more relevant as touring income has come to a halt for musicians and the live entertainment industry. <br/>Under the current per-stream model, it is becoming exceedingly hard for artists to make a living off of streams. This forces artists to tour heavily as well as cut corners to create what is essentially “disposable art”. Rapidly releasing multiple projects a year has become the norm for many modern artists. This paper will examine the licensing framework, royalty payout issues, and propose a solution.
ContributorsKoudssi, Zakaria Corley (Author) / Sadusky, Brian (Thesis director) / Koretz, Lora (Committee member) / Dean, W.P. Carey School of Business (Contributor) / Department of Finance (Contributor) / Barrett, The Honors College (Contributor)
Created2021-05
Description
The aim of this project was to provide college applicants with the ability to apply using a video instead of an essay. These videos are analyzed automatically and their scripts are taken and submitted with the application. This was implemented through the use of Amazon Web Services (AWS) and their S3 buckets along with their speech to text transcription service. This type of application process can give admissions teams the opportunity to get to know who will potentially be attending their university and allows the applicants to express themselves to admissions teams in a new and unique way.
ContributorsStephan, Meagan (Co-author) / Pratt, Devan (Co-author) / Chen, Yinong (Thesis director) / Balasooriya, Janaka (Committee member) / Computer Science and Engineering Program (Contributor) / Barrett, The Honors College (Contributor)
Created2019-12
Description
Although Spotify’s extensive library of songs are often seen broken up by “Top 100” and main lyrical genres, these categories are primarily based on popularity, artist and general mood alone. If a user wanted to create a playlist based on specific or situationally specific qualifiers from their own downloaded library, he/she would have to hand pick songs that fit the mold and create a new playlist. This is a time consuming process that may not produce the most efficient result due to human error. The objective of this project, therefore, was to develop an application to streamline this process, optimize efficiency, and fill this user need.
Song Sift is an application built using Angular that allows users to filter and sort their song library to create specific playlists using the Spotify Web API. Utilizing the audio feature data that Spotify attaches to every song in their library, users can filter their downloaded Spotify songs based on four main attributes: (1) energy (how energetic a song sounds), (2) danceability (how danceable a song is), (3) valence (how happy a song sounds), and (4) loudness (average volume of a song). Once the user has created a playlist that fits their desired genre, he/she can easily export it to their Spotify account with the click of a button.
Song Sift is an application built using Angular that allows users to filter and sort their song library to create specific playlists using the Spotify Web API. Utilizing the audio feature data that Spotify attaches to every song in their library, users can filter their downloaded Spotify songs based on four main attributes: (1) energy (how energetic a song sounds), (2) danceability (how danceable a song is), (3) valence (how happy a song sounds), and (4) loudness (average volume of a song). Once the user has created a playlist that fits their desired genre, he/she can easily export it to their Spotify account with the click of a button.
ContributorsDiMuro, Louis (Author) / Balasooriya, Janaka (Thesis director) / Chen, Yinong (Committee member) / Arts, Media and Engineering Sch T (Contributor) / Computer Science and Engineering Program (Contributor) / Barrett, The Honors College (Contributor)
Created2020-05