2023-06-15: A Milestone Achieved: Completing my Master's Degree and Advancing to PhD Candidacy

In the Fall of 2019, I joined the WS-DL (Web Science and Digital Libraries) research group at Old Dominion University as a Master’s student under the supervision of Dr. Michele C. Weigle and Dr. Michael L. Nelson. During my master's degree, I was on the thesis option (24 coursework credits, 6 research credits, a thesis, and a thesis defense). As I progressed through my courses and research, I discovered my passion for studying and research and  I realized that I wanted to continue my academic journey and pursue a PhD. I reached out to my MS thesis advisor, Dr. Weigle, to express my strong interest in pursuing a PhD within the exceptional WS-DL group. I applied to the PhD program at ODU and was thrilled to be accepted for the Fall 2020 semester. Since then, I have been a dual-status MS and PhD student, I have been working on completing my Master's thesis while also advancing my PhD studies.

Figure 1:Some cherished memories with my incredible support system: advisors, mentors, peers, and my family. 

My master’s thesis is titled "Supporting Account-based Queries for Archived Instagram Posts". This topic crossed our path during our preliminary research on Instagram content in web archives, fueled by our interest in the platform. We conducted several early studies, such as "How well is Instagram archived?" which evolved from a class project in Dr. Nelson's Web Archiving Forensics course, and "Creation Time and Published Time Are Not the Same: Estimating the Instagram Epoch" aimed at estimating the Instagram epoch value. Our research on Twitter web archiving had been extensive, and we observed how comparatively under-studied Instagram was. During one of my weekly meetings with Drs. Weigle and Nelson, we discovered that the post owner's username is not included in the post URL on Instagram. Figure 2 illustrates the distinct difference in URL structure among the three social media platforms: Facebook, Twitter, and Instagram. While Facebook and Twitter include the usernames of post owners in their individual post URLs, Instagram does not incorporate usernames in its URLs. This is the root cause of the problem I have explored in the thesis.

Figure 2: While usernames are retained on Facebook and Twitter for individual posts, they are not available on Instagram.

The most popular method for searching for content in web archives like the Internet Archive is to use the URL of a particular web resource as the lookup key in the search bar. An archived web page, or memento (URI-M), is a snapshot of an original resource (URI-R) captured by a web archive at a fixed moment in time (Memento-Datetime). A list of URIs for mementos of the original resource is referred to as the TimeMap (URI-T). As a result of not knowing the URL of the resource (post URL), there is a lot of Instagram web content preserved in web archives that the user is unable to discover. The CDX is a type of index used in the field of web archiving that provides a simple representation of metadata from all records in an archive. CDX files are created as an index of generated WARC files. The CDX server can be used to list every URI-M in the Internet Archive index for a given URI-R. We can use the CDX API to return results matching a specific “prefix”'. For Facebook, a CDX prefix query http://web.archive.org/cdx/?url=facebook.com/{username}/posts/&matchType=prefix would provide the web archive user with all the available Facebook status posts that belong to a particular user (Figure 3). Similarly for Twitter, a simple CDX prefix query http://web.archive.org/cdx/?url=twitter.com/{username}/status/&matchType=prefix would help a web archive user to identify all the available Twitter status posts that belong to a particular user (Figure 4). However, there is no left-to-right hierarchical construction of URLs on Instagram. There is no way to query for URI-Ms of all of the posts associated with a specific Instagram account without knowing their URLs. 

$ curl -s "http://web.archive.org/cdx/search/cdx?url=https://www.facebook.com/katyperry/posts/&matchType=prefix" | head -2

com,facebook)/katyperry/posts/10150201275371466 20160305122514 http://www.facebook.com/katyperry/posts/10150201275371466 text/html 302 3I42H3S6NNFQ2MSVX7XZKYAYSCX5QBYJ 492

com,facebook)/katyperry/posts/10150295888246466 20111015052123 http://www.facebook.com/katyperry/posts/10150295888246466 text/html 200 TPOUCAFF7PUU34UHMYJCY42KI66HRPSH 13489

Figure 3: A CDX “Prefix” Query for Facebook Posts

$ curl -s "http://web.archive.org/cdx/search/cdx?url=https://twitter.com/katyperry/status/&matchType=prefix" | head -1944 | tail -2

com,twitter)/katyperry/status/1068587128974524416 20211130163200 https://twitter.com/katyperry/status/1068587128974524416 text/html 200 DDX4LIM5MF5XSEEMLJVZOW4S5TW7DACL 8799

com,twitter)/katyperry/status/1068608473913409536 20181130221008 https://twitter.com/katyperry/status/1068608473913409536 text/html 200 JLB2P6ECZL23CCFCKXLMBGQNCVQ4F3M4 64450

Figure 4: A CDX “Prefix” Query for Twitter Posts (Tweets)

To solve this problem, we proposed two approaches.

1. The first approach uses existing technologies in the Internet Archive by using WARC revisit records to incorporate Instagram usernames into the WARC-Target-URI field in the WARC file header. The process of using revisit records to automatically enable prefix search for discovering Instagram posts belonging to a particular account is illustrated in Figure 5.

2. The second approach involves building an external index that maps Instagram user accounts to their posts. The user can query this index to retrieve all post URLs for a particular user, which they can then use to query web archives for each individual post. The techniques that we are proposing to fill the index are summarized in Figure 6.


Figure 5: Summary of the process involved in demonstrating prefix search for Instagram top-level post queries through the use of revisit records.

Figure 6: The techniques that can be used to populate the proposed secondary index. 

Our implementation of various approaches has successfully demonstrated facilitating the discovery of archived Instagram posts belonging to a specific user. 

Other interesting findings:

* By analyzing the server access logs of the Internet Archive from 2011 to 2021 (the first Thursday of February each year), we observed a growing trend in the number of requests for the Instagram domain, particularly until 2020 (Figure 7). However, the percentage of the number of requests out of all requests remained below 0.2% each year. 

Figure 7:  The number of requests for the Instagram domain increased over the years, at least until 2020, but the percentage of requests for Instagram is still less than 0.2% each year compared to the total number of raw access logs.

* The number of requested posts has consistently increased from 2013 to 2021, with a significant surge in 2021 compared to 2020 (Figure 8). 

Figure 8: There is a steady increase in the number of posts requested each year from 2013 to 2021, with a notable surge in 2021.

* The majority of web archive users request mementos of Instagram posts that are created recently. For the years 2013 to 2017 (Figure 9 - left), 50% of the requests were for posts created within one year prior to the access logs' datetime. For the years 2018 to 2021 (Figure 9 - right), 50% of requests were for posts created within two years prior to the access logs' datetime. 

Figure 9: The majority of web archive users request mementos of Instagram posts that are created recently.

On April 13, 2023, I successfully defended my MS thesis. I have embedded the defense recording and the slides used in my defense presentation below. The full thesis document is available on ODU Digital Commons.

I am grateful to Drs. Michele C. Weigle and Michael L. Nelson for being my advisors during my master's program. Their expert guidance and unwavering support proved invaluable in facilitating the entire process, from the initial stages of selecting a topic to the final outcome. During our weekly meetings, we engaged in numerous discussions and brainstorming sessions, which were immensely helpful for me. I express my gratitude to my advisors and Dr. Faryaneh Poursardar for serving on my thesis committee and for their insightful feedback throughout the entire process.

Moreover, I am delighted to share that my master's thesis defense presentation and thesis document got approved as my PhD candidacy exam. This accomplishment brought me immense satisfaction in successfully fulfilling both requirements: officially completing my master's degree and passing the PhD candidacy exam. I also think that this milestone is a good time to reflect on my academic journey so far (listed towards the end of this blog post), especially since many people seem to be confused and think I just completed my PhD instead of my Master's degree. But on the bright side, it's always good to be ahead of the game, right? In all seriousness, despite taking longer than the typical duration for an MS degree, I am proud of my achievements, and it has only strengthened my resolve to continue my academic pursuits. Additionally, I hope to encourage others interested in pursuing graduate studies that timelines are not set in stone and as long as the work is done diligently, there is ample flexibility to navigate through this exciting journey. I would like to express my gratitude once again for the understanding and support of my advisors throughout my studies. Thank you for believing in me and for continuously pushing me to be my best.

What lies ahead for me? My next step is to shift my focus toward my doctoral dissertation. The knowledge and skills I have acquired through working on multiple research projects, completing my MS thesis, and publishing papers have only strengthened my desire to further engage in the academic world of learning and research. I am beyond excited to continue my doctoral research with the WS-DL research group, where the perfect blend of excellence, experience, and infectious awesomeness of faculty and students awaits!

-- Himarsha R. Jayanetti


Semester Highlights: Projects, Publications, Blogs, and Awards

Since I enrolled in the MS program in the Fall of 2019, I have accomplished the following academic milestones:

* Fall 2019: Joined ODU as an MS student in Computer Science, pursuing the MS thesis option.

* Spring 2020: I began analyzing web archive server access logs with Kritika Garg, under the guidance of WS-DL alumnus Dr. Sawood Alam, who is currently working as a Web & Data Scientist at the Internet Archive (my first ever research publication as first author - full paper at TPDL 2022, arXiv pre-print, 🏆 Best Student Paper Award).

* Summer 2020: I collaborated with Muntabir Choudhury under the supervision of Dr. Wu, focusing on mining metadata from scanned ETDs (short paper at JCDL 2021, arXiv pre-print). I also started working with Kritika Garg on exploring challenges related to archiving Twitter (full paper at JCDL 2021, arXiv pre-print).

* Fall 2020: I enrolled in the ODU Computer Science PhD program under the guidance of my MS thesis advisor Dr. Weigle, who remains my advisor.

* Spring, Summer, and Fall 2021: I worked on the Dark & Stormy Archives project, which was funded by the IIPC. Working alongside project lead Shawn M. Jones, who is a WS-DL alumnus and current ISTI Postdoctoral Fellow at Los Alamos National Laboratory’s Information Sciences Division, was an invaluable experience. (short paper at TPDL 2022, arXiv pre-print, another article in the Code4Lib Journal)

* Spring 2022: I worked again under the supervision of Dr. Wu, focusing on ETD data enhancement and quality improvement. In this project, I had the chance to collaborate with Muntabir Choudhury and Lamia Salsabil (short paper at JCDL 2023, arXiv pre-print, 🏆 Nominated for Best Short Paper). I was honored to receive the Dr. Irwin B. Levinstein Scholarship for graduate students based on their academic progress and accomplishments in the Computer Science department at ODU. At the end of Spring 2022, I completed the MS course requirements (including the 1-credit colloquium) and the PhD breadth course requirements. 

* Summer 2022: I interned at the Los Alamos National Laboratory (LANL) under the Graduate Research Assistant Program. I was under the supervision of Dr. Shawn M. Jones and Dr. Martin Klein as a part of the Library’s Research and Prototyping Team (Proto Team) where I developed a social card generator for scientific PDF documents. This is one of the three components of the PDFServer developed by me and two other LANL summer interns: Gavindya Jayawardena and Yasith Jayawardana (my first summer internship as a graduate student). I also got the opportunity to help out with the summer REU program and work with Haley Bragg (REU student) from CNU and Dr. Weigle on their project “Discovering the traces of disinformation on Instagram”. (short paper at JCDL 2023, 🏆 Nominated for Best Short Paper)

* Fall 2022 (continued to Spring and Summer of 2023): I am working on the project "Data Science for Social Good: Mining and Visualizing Worldwide News to Monitor Xenophobic Violence" under the supervision of Dr. Erika Frydenlund, a Research Assistant Professor at VMASC and Dr. Weigle. This project is funded by the 2022-2023 ODU Data Science Seed Funding Program. (full paper at MSVSCC 2023, arXiv pre-print, 🏆 Best Overall Paper and 🏆Best Paper in Data Science Track Award)

* Spring 2023: On April 13, 2023, I defended my master's thesis, a significant milestone in my academic journey. On May 4, 2023, I received the incredible news that my master's thesis document and defense were both approved as my PhD candidacy, officially making me a PhD candidate. On May 5, 2023, I had the privilege of participating in the commencement ceremony for my master's degree. The celebration of my academic journey so far and a deep sense of success made it an experience I'll never forget.

Figure 10: My avatar on the PhD crush board of our group reflecting the progress in WS-DL style ✨

Throughout the period spanning from the Fall of 2019 to the present, I published several research papers. Specifically, I have contributed 7 conference papers, 2 workshop papers, and 2 posters to the scholarly community. Additionally, there are two journal papers that are under review.

I have also produced 12 blog posts in the WS-DL blog:

1. 2020-01-01: Himarsha Jayanetti (Computer Science Master’s Student) - My welcome blog post for the WS-DL group
9. 2021-09-20: Digging Up a Gem Through the Web Archives - Thanks to a tweet by the Internet Archive and a retweet of my post by the Wayback Machine, this blog has become the most viewed among all, gaining over 1000 views.

12. 2022-12-21: PDFServer - Our Summer Internship at LANL - with Yasith Jayawardana and Gavindya Jayawardena

Additionally, our research group acknowledges student callouts by name on the DSHR’s blog while at ODU, and I was honored to receive two mentions alongside Kritika for our work on archiving Twitter.

1. 2020-12-29: Michael Nelson's Group On Archiving Twitter

2. 2021-02-11: More On Archiving Twitter 

I would like to express my deepest appreciation to my collaborators, as well as the entire WS-DL group, including both current members and esteemed alumni (WS-DL for life!). I would also like to express my heartfelt gratitude to my advisors, internship supervisors, and all faculty members for their invaluable guidance and the opportunities they extend to students.