Michael L. Nelson

Ahmed AlSum, Michael L. Nelson

Archiving the web is socially and culturally critical, but presents problems of scale. The Internet Archive's Wayback Machine can replay captured web pages as they existed at a certain point in time, but it has limited ability to provide extensive content and structural metadata about the web graph. While the live web has developed a rich ecosystem of APIs to facilitate web applications (e.g., APIs from Google and Twitter), the web archiving community has not yet broadly implemented this level of access. We present ArcLink, a proof-of-concept system that complements open source Wayback Machine installations by optimizing the construction, storage, and access to the temporal web graph. We divide the web graph construction into four stages (filtering, extraction, storage, and access) and explore optimization for each stage. ArcLink extends the current Web archive interfaces to return content and structural metadata for each URI. We show how this API can be applied to such applications as retrieving inlinks, outlinks, anchortext, and PageRank.

Sawood Alam, Michele C. Weigle, Michael L. Nelson

Prior work on web archive profiling were focused on Archival Holdings to describe what is present in an archive. This work defines and explores Archival Voids to establish a means to represent portions of URI spaces that are not present in a web archive. Archival Holdings and Archival Voids profiles can work independently or as complements to each other to maximize the Accuracy of Memento Aggregators. We discuss various sources of truth that can be used to create Archival Voids profiles. We use access logs from Arquivo.pt to create various Archival Voids profiles and analyze them against our MemGator access logs for evaluation. We find that we could have avoided more than 8% of additional False Positives on top of the 60% Accuracy we got from profiling Archival Holdings in our prior work, if Arquivo.pt were to provide an Archival Voids profile based on URIs that were requested hundreds of times and never returned any success responses.

Shawn M. Jones, Michele C. Weigle, Martin Klein et al.

To allow previewing a web page, social media platforms have developed social cards: visualizations consisting of vital information about the underlying resource. At a minimum, social cards often include features such as the web resource's title, text summary, striking image, and domain name. News and scholarly articles on the web are frequently subject to social card creation when being shared on social media. However, we noticed that not all web resources offer sufficient metadata elements to enable appealing social cards. For example, the COVID-19 emergency has made it clear that scholarly articles, in particular, are at an aesthetic disadvantage in social media platforms when compared to their often more flashy disinformation rivals. Also, social cards are often not generated correctly for archived web resources, including pages that lack or predate standards for specifying striking images. With these observations, we are motivated to quantify the levels of inclusion of required metadata in web resources, its evolution over time for archived resources, and create and evaluate an algorithm to automatically select a striking image for social cards. We find that more than 40% of archived news articles sampled from the NEWSROOM dataset and 22% of scholarly articles sampled from the PubMed Central dataset fail to supply striking images. We demonstrate that we can automatically predict the striking image with a Precision@1 of 0.83 for news articles from NEWSROOM and 0.78 for scholarly articles from the open access journal PLOS ONE.

Yasith Jayawardana, Alexander C. Nwala, Gavindya Jayawardena et al.

The vastness of the web imposes a prohibitive cost on building large-scale search engines with limited resources. Crawl frontiers thus need to be optimized to improve the coverage and freshness of crawled content. In this paper, we propose an approach for modeling the dynamics of change in the web using archived copies of webpages. To evaluate its utility, we conduct a preliminary study on the scholarly web using 19,977 seed URLs of authors' homepages obtained from their Google Scholar profiles. We first obtain archived copies of these webpages from the Internet Archive (IA), and estimate when their actual updates occurred. Next, we apply maximum likelihood to estimate their mean update frequency ($λ$) values. Our evaluation shows that $λ$ values derived from a short history of archived data provide a good estimate for the true update frequency in the short-term, and that our method provides better estimations of updates at a fraction of resources compared to the baseline models. Based on this, we demonstrate the utility of archived data to optimize the crawling strategy of web crawlers, and uncover important challenges that inspire future research directions.

Shawn M. Jones, Alexander C. Nwala, Martin Klein et al.

Tools such as Google News and Flipboard exist to convey daily news, but what about the past? In this paper, we describe how to combine several existing tools with web archive holdings to perform news analysis and visualization of the "biggest story" for a given date. StoryGraph clusters news articles together to identify a common news story. Hypercane leverages ArchiveNow to store URLs produced by StoryGraph in web archives. Hypercane analyzes these URLs to identify the most common terms, entities, and highest quality images for social media storytelling. Raintale then uses the output of these tools to produce a visualization of the news story for a given day. We name this process SHARI (StoryGraph Hypercane ArchiveNow Raintale Integration).

Shawn M. Jones, Martin Klein, Michele C. Weigle et al.

For traditional library collections, archivists can select a representative sample from a collection and display it in a featured physical or digital library space. Web archive collections may consist of thousands of archived pages, or mementos. How should an archivist display this sample to drive visitors to their collection? Search engines and social media platforms often represent web pages as cards consisting of text snippets, titles, and images. Web storytelling is a popular method for grouping these cards in order to summarize a topic. Unfortunately, social media platforms are not archive-aware and fail to consistently create a good experience for mementos. They also allow no UI alterations for their cards. Thus, we created MementoEmbed to generate cards for individual mementos and Raintale for creating entire stories that archivists can export to a variety of formats.

Alexander C. Nwala, Michele C. Weigle, Michael L. Nelson

We investigate the overlap of topics of online news articles from a variety of sources. To do this, we provide a platform for studying the news by measuring this overlap and scoring news stories according to the degree of attention in near-real time. This can enable multiple studies, including identifying topics that receive the most attention from news organizations and identifying slow news days versus major news days. Our application, StoryGraph, periodically (10-minute intervals) extracts the first five news articles from the RSS feeds of 17 US news media organizations across the partisanship spectrum (left, center, and right). From these articles, StoryGraph extracts named entities (PEOPLE, LOCATIONS, ORGANIZATIONS, etc.) and then represents each news article with its set of extracted named entities. Finally, StoryGraph generates a news similarity graph where the nodes represent news articles, and an edge between a pair of nodes represents a high degree of similarity between the nodes (similar news stories). Each news story within the news similarity graph is assigned an attention score which quantifies the amount of attention the topics in the news story receive collectively from the news media organizations. The StoryGraph service has been running since August 2017, and using this method, we determined that the top news story of 2018 was the "Kavanaugh hearings" with attention score of 25.85 on September 27, 2018. Similarly, the top news story for 2019 so far (2019-12-12) is "AG William Barr's release of his principal conclusions of the Mueller Report," with an attention score of 22.93 on March 24, 2019.

Lulwah M. Alkwai, Michael L. Nelson, Michele C. Weigle

When a user requests a web page from a web archive, the user will typically either get an HTTP 200 if the page is available, or an HTTP 404 if the web page has not been archived. This is because web archives are typically accessed by URI lookup, and the response is binary: the archive either has the page or it does not, and the user will not know of other archived web pages that exist and are potentially similar to the requested web page. In this paper, we propose augmenting these binary responses with a model for selecting and ranking recommended web pages in a Web archive. This is to enhance both HTTP 404 responses and HTTP 200 responses by surfacing web pages in the archive that the user may not know existed. First, we check if the URI is already classified in DMOZ or Wikipedia. If the requested URI is not found, we use ML to classify the URI using DMOZ as our ontology and collect candidate URIs to recommended to the user. Next, we filter the candidates based on if they are present in the archive. Finally, we rank candidates based on several features, such as archival quality, web page popularity, temporal similarity, and URI similarity. We calculated the F1 score for different methods of classifying the requested web page at the first level. We found that using all-grams from the URI after removing numerals and the TLD produced the best result with F1=0.59. For second-level classification, the micro-average F1=0.30. We found that 44.89% of the correctly classified URIs contained at least one word that exists in a dictionary and 50.07% of the correctly classified URIs contained long strings in the domain. In comparison with the URIs from our Wayback access logs, only 5.39% of those URIs contained only words from a dictionary, and 26.74% contained at least one word from a dictionary. These percentages are low and may affect the ability for the requested URI to be correctly classified.

Sawood Alam, Michele C. Weigle, Michael L. Nelson et al.

We describe challenges related to web archiving, replaying archived web resources, and verifying their authenticity. We show that Web Packaging has significant potential to help address these challenges and identify areas in which changes are needed in order to fully realize that potential.

Alexander C. Nwala, Michele C. Weigle, Michael L. Nelson

In a Web plagued by disappearing resources, Web archive collections provide a valuable means of preserving Web resources important to the study of past events ranging from elections to disease outbreaks. These archived collections start with seed URIs (Uniform Resource Identifiers) hand-selected by curators. Curators produce high quality seeds by removing non-relevant URIs and adding URIs from credible and authoritative sources, but it is time consuming to collect these seeds. Two main strategies adopted by curators for discovering seeds include scraping Web (e.g., Google) Search Engine Result Pages (SERPs) and social media (e.g., Twitter) SERPs. In this work, we studied three social media platforms in order to provide insight on the characteristics of seeds generated from different sources. First, we developed a simple vocabulary for describing social media posts across different platforms. Second, we introduced a novel source for generating seeds from URIs in the threaded conversations of social media posts created by single or multiple users. Users on social media sites routinely create and share posts about news events consisting of hand-selected URIs of news stories, tweets, videos, etc. In this work, we call these posts micro-collections, and we consider them as an important source for seeds because the effort taken to create micro-collections is an indication of editorial activity, and a demonstration of domain expertise. Third, we generated 23,112 seed collections with text and hashtag queries from 449,347 social media posts from Reddit, Twitter, and Scoop.it. We collected in total 120,444 URIs from the conventional scraped SERP posts and micro-collections. We characterized the resultant seed collections across multiple dimensions including the distribution of URIs, precision, ages, diversity of webpages, etc...

Shawn M. Jones, Michele C. Weigle, Michael L. Nelson

Used by a variety of researchers, web archive collections have become invaluable sources of evidence. If a researcher is presented with a web archive collection that they did not create, how do they know what is inside so that they can use it for their own research? Search engine results and social media links are represented as surrogates, small easily digestible summaries of the underlying page. Search engines and social media have a different focus, and hence produce different surrogates than web archives. Search engine surrogates help a user answer the question "Will this link meet my information need?" Social media surrogates help a user decide "Should I click on this?" Our use case is subtly different. We hypothesize that groups of surrogates together are useful for summarizing a collection. We want to help users answer the question of "What does the underlying collection contain?" But which surrogate should we use? With Mechanical Turk participants, we evaluate six different surrogate types against each other. We find that the type of surrogate does not influence the time to complete the task we presented the participants. Of particular interest are social cards, surrogates typically found on social media, and browser thumbnails, screen captures of web pages rendered in a browser. At $p=0.0569$, and $p=0.0770$, respectively, we find that social cards and social cards paired side-by-side with browser thumbnails probably provide better collection understanding than the surrogates currently used by the popular Archive-It web archiving platform. We measure user interactions with each surrogate and find that users interact with social cards less than other types. The results of this study have implications for our web archive summarization work, live web curation platforms, social media, and more.

Grant C. Atkins, Alexander Nwala, Michele C. Weigle et al.

News websites make editorial decisions about what stories to include on their website homepages and what stories to emphasize (e.g., large font size for main story). The emphasized stories on a news website are often highly similar to many other news websites (e.g, a terrorist event story). The selective emphasis of a top news story and the similarity of news across different news organizations are well-known phenomena but not well-measured. We provide a method for identifying the top news story for a select set of U.S.-based news websites and then quantify the similarity across them. To achieve this, we first developed a headline and link extractor that parses select websites, and then examined ten United States based news website homepages during a three month period, November 2016 to January 2017. Using archived copies, retrieved from the Internet Archive (IA), we discuss the methods and difficulties for parsing these websites, and how events such as a presidential election can lead news websites to alter their document representation just for these events. We use our parser to extract k = 1, 3, 10 maximum number of stories for each news site. Second, we used the cosine similarity measure to calculate news similarity at 8PM Eastern Time for each day in the three months. The similarity scores show a buildup (0.335) before Election Day, with a declining value (0.328) on Election Day, and an increase (0.354) after Election Day. Our method shows that we can effectively identity top stories and quantify news similarity.

Shawn M. Jones, Michele C. Weigle, Michael L. Nelson

Web archive collections are created with a particular purpose in mind. A curator selects seeds, or original resources, which are then captured by an archiving system and stored as archived web pages, or mementos. The systems that build web archive collections are often configured to revisit the same original resource multiple times. This is incredibly useful for understanding an unfolding news story or the evolution of an organization. Unfortunately, over time, some of these original resources can go off-topic and no longer suit the purpose for which the collection was originally created. They can go off-topic due to web site redesigns, changes in domain ownership, financial issues, hacking, technical problems, or because their content has moved on from the original topic. Even though they are off-topic, the archiving system will still capture them, thus it becomes imperative to anyone performing research on these collections to identify these off-topic mementos. Hence, we present the Off-Topic Memento Toolkit, which allows users to detect off-topic mementos within web archive collections. The mementos identified by this toolkit can then be separately removed from a collection or merely excluded from downstream analysis. The following similarity measures are available: byte count, word count, cosine similarity, Jaccard distance, Sørensen-Dice distance, Simhash using raw text content, Simhash using term frequency, and Latent Semantic Indexing via the gensim library. We document the implementation of each of these similarity measures. We possess a gold standard dataset generated by manual analysis, which contains both off-topic and on-topic mementos. Using this gold standard dataset, we establish a default threshold corresponding to the best F1 score for each measure. We also provide an overview of potential future directions that the toolkit may take.

Justin F. Brunelle, Michele C. Weigle, Michael L. Nelson

Web resources are increasingly interactive, resulting in resources that are increasingly difficult to archive. The archival difficulty is based on the use of client-side technologies (e.g., JavaScript) to change the client-side state of a representation after it has initially loaded. We refer to these representations as deferred representations. We can better archive deferred representations using tools like headless browsing clients. We use 10,000 seed Universal Resource Identifiers (URIs) to explore the impact of including PhantomJS -- a headless browsing tool -- into the crawling process by comparing the performance of wget (the baseline), PhantomJS, and Heritrix. Heritrix crawled 2.065 URIs per second, 12.15 times faster than PhantomJS and 2.4 times faster than wget. However, PhantomJS discovered 531,484 URIs, 1.75 times more than Heritrix and 4.11 times more than wget. To take advantage of the performance benefits of Heritrix and the URI discovery of PhantomJS, we recommend a tiered crawling strategy in which a classifier predicts whether a representation will be deferred or not, and only resources with deferred representations are crawled with PhantomJS while resources without deferred representations are crawled with Heritrix. We show that this approach is 5.2 times faster than using only PhantomJS and creates a frontier (set of URIs to be crawled) 1.8 times larger than using only Heritrix.

Sawood Alam, Fateh ud din B Mehmood, Michael L. Nelson

We propose an approach to index raster images of dictionary pages which in turn would require very little manual effort to enable direct access to the appropriate pages of the dictionary for lookup. Accessibility is further improved by feedback and crowdsourcing that enables highlighting of the specific location on the page where the lookup word is found, annotation, digitization, and fielded searching. This approach is equally applicable on simple scripts as well as complex writing systems. Using our proposed approach, we have built a Web application called "Dictionary Explorer" which supports word indexes in various languages and every language can have multiple dictionaries associated with it. Word lookup gives direct access to appropriate pages of all the dictionaries of that language simultaneously. The application has exploration features like searching, pagination, and navigating the word index through a tree-like interface. The application also supports feedback, annotation, and digitization features. Apart from the scanned images, "Dictionary Explorer" aggregates results from various sources and user contributions in Unicode. We have evaluated the time required for indexing dictionaries of different sizes and complexities in the Urdu language and examined various trade-offs in our implementation. Using our approach, a single person can make a dictionary of 1,000 pages searchable in less than an hour.

Yasmin AlNoamany, Michele C. Weigle, Michael L. Nelson

Although user access patterns on the live web are well-understood, there has been no corresponding study of how users, both humans and robots, access web archives. Based on samples from the Internet Archive's public Wayback Machine, we propose a set of basic usage patterns: Dip (a single access), Slide (the same page at different archive times), Dive (different pages at approximately the same archive time), and Skim (lists of what pages are archived, i.e., TimeMaps). Robots are limited almost exclusively to Dips and Skims, but human accesses are more varied between all four types. Robots outnumber humans 10:1 in terms of sessions, 5:4 in terms of raw HTTP accesses, and 4:1 in terms of megabytes transferred. Robots almost always access TimeMaps (95% of accesses), but humans predominately access the archived web pages themselves (82% of accesses). In terms of unique archived web pages, there is no overall preference for a particular time, but the recent past (within the last year) shows significant repeat accesses.

Hany M. SalahEldeen, Michael L. Nelson

In previous work we reported that resources linked in tweets disappeared at the rate of 11% in the first year followed by 7.3% each year afterwards. We also found that in the first year 6.7%, and 14.6% in each subsequent year, of the resources were archived in public web archives. In this paper we revisit the same dataset of tweets and find that our prior model still holds and the calculated error for estimating percentages missing was about 4%, but we found the rate of archiving produced a higher error of about 11.5%. We also discovered that resources have disappeared from the archives themselves (7.89%) as well as reappeared on the live web after being declared missing (6.54%). We have also tested the availability of the tweets themselves and found that 10.34% have disappeared from the live web. To mitigate the loss of resources on the live web, we propose the use of a "tweet signature". Using the Topsy API, we extract the top five most frequent terms from the union of all tweets about a resource, and use these five terms as a query to Google. We found that using tweet signatures results in discovering replacement resources with 70+% textual similarity to the missing resource 41% of the time.

Hany M. SalahEldeen, Michael L. Nelson

The web is trapped in the "perpetual now", and when users traverse from page to page, they are seeing the state of the web resource (i.e., the page) as it exists at the time of the click and not necessarily at the time when the link was made. Thus, a temporal discrepancy can arise between the resource at the time the page author created a link to it and the time when a reader follows the link. This is especially important in the context of social media: the ease of sharing links in a tweet or Facebook post allows many people to author web content, but the space constraints combined with poor awareness by authors often prevents sufficient context from being generated to determine the intent of the post. If the links are clicked as soon as they are shared, the temporal distance between sharing and clicking is so small that there is little to no difference in content. However, not all clicks occur immediately, and a delay of days or even hours can result in reading something other than what the author intended. We introduce the concept of a user's temporal intention upon publishing a link in social media. We investigate the features that could be extracted from the post, the linked resource, and the patterns of social dissemination to model this user intention. Finally, we analyze the historical integrity of the shared resources in social media across time. In other words, how much is the knowledge of the author's intent beneficial in maintaining the consistency of the story being told through social posts and in enriching the archived content coverage and depth of vulnerable resources?

Sawood Alam, Charles L. Cartledge, Michael L. Nelson

We describe HTTP Mailbox, a mechanism to enable RESTful HTTP communication in an asynchronous mode with a full range of HTTP methods otherwise unavailable to standard clients and servers. HTTP Mailbox allows for broadcast and multicast semantics via HTTP. We evaluate a reference implementation using ApacheBench (a server stress testing tool) demonstrating high throughput (on 1,000 concurrent requests) and a systemic error rate of 0.01%. Finally, we demonstrate our HTTP Mailbox implementation in a human assisted web preservation application called "Preserve Me".

Hany M. SalahEldeen, Michael L. Nelson

In the course of web research it is often necessary to estimate the creation datetime for web resources (in the general case, this value can only be estimated). While it is feasible to manually establish likely datetime values for small numbers of resources, this becomes infeasible if the collection is large. We present "carbon date", a simple web application that estimates the creation date for a URI by polling a number of sources of evidence and returning a machine-readable structure with their respective values. To establish a likely datetime, we poll bitly for the first time someone shortened the URI, topsy for the first time someone tweeted the URI, a Memento aggregator for the first time it appeared in a public web archive, Google's time of last crawl, and the Last-Modified HTTP response header of the resource itself. We also examine the backlinks of the URI as reported by Google and apply the same techniques for the resources that link to the URI. We evaluated our tool on a gold-standard data set of 1200 URIs in which the creation date was manually verified. We were able to estimate a creation date for 75.90% of the resources, with 32.78% having the correct value. Given the different nature of the URIs, the union of the various methods produces the best results. While the Google last crawl date and topsy account for nearly 66% of the closest answers, eliminating the web archives or Last-Modified from the results produces the largest overall negative impact on the results. The carbon date application is available for download or use via a webAPI.

Hany M. SalahEldeen, Michael L. Nelson

Social media content has grown exponentially in the recent years and the role of social media has evolved from just narrating life events to actually shaping them. In this paper we explore how many resources shared in social media are still available on the live web or in public web archives. By analyzing six different event-centric datasets of resources shared in social media in the period from June 2009 to March 2012, we found about 11% lost and 20% archived after just a year and an average of 27% lost and 41% archived after two and a half years. Furthermore, we found a nearly linear relationship between time of sharing of the resource and the percentage lost, with a slightly less linear relationship between time of sharing and archiving coverage of the resource. From this model we conclude that after the first year of publishing, nearly 11% of shared resources will be lost and after that we will continue to lose 0.02% per day.