Searching on the Web has become an essential part of learning, and bookmarking is a way to remember relevant and interesting sites. Even though bookmarking systems have been around since the dawn of the Web, they have not evolved much in the last 20 years. In this paper, we introduce LogCanvas v2, a new and extended design of the search history interface, intended to capture the search and learning process in our Learnweb educational platform. Our new interface focuses on users' queries rather than just the browsed webpages, enabling users to reconstruct the searching, browsing and learning process. Also, it helps them to re-find the information they need and annotate the useful information. We hypothesize that search learning with annotation capabilities can be achieved in our platform. We finish the paper with a detailed description of a learning scenario and the benefits of the learning process in our platform.
the user experience in searching to learn scenarios, when users want to revisit certain webpages to review what Searching to learn is increasingly viable as more ma- they have learned during the searching process? To anterials become digital and get published on the web. swer this question, we carried out a comparative study Learning searches involve multiple iterations of queries of user experience in using history interfaces of popuand return sets of objects that require cognitive pro- lar search engines to review webpages while searching cessing and interpretation. During this process, sear- to learn (see Section 3). Based on the results of this inchers’ interactions with search systems is generally vestigations, we propose a new design for the search recorded as search history logs. Studies found that as engine history interface, the LogCanvas v2. In conmany of 40% of users’ search queries are attempts to trast to the existing interfaces, the LogCanvas v2 fore-find previously encountered results [1]. Further, a cuses more on the users’ searching and browsing path, survey of experienced Web users showed that people such as how users issued queries and navigated to cerwould like to use search engines to re-find online in- tain search results; which enables them to reconstruct formation, but often have dificulty remembering the their searching, browsing and learning process to help sequence of queries they had used when they origi- them re-find information quickly. nally discovered the content in question [2]. In this This paper is organized as follows: in Section 2, we scenario, search history can be an important resource provide an overview of related literature about existfor both individuals and collaborative searching groups ing search history visualization platforms. In Section 3, to preserve and recall their searching and learning pro- we discuss users’ experience in using current search cess. engines’ history interfaces and their feedback. In Sec
Search logs record explicit activities of searchers, tion 4, we describe our design and the workflow of the including the submitted queries and the clicked an- LogCanvas v2 history interface. An experimental sceswers (search results). The history interface in cur- nario in which we plan to evaluate Learnweb is prerent search engines displays the issued queries with sented in Section 5. Finally, we draw our conclusions corresponding visited webpages in the form of a URL in Section 6. list. However, there is a research question that might arise: is the current history interface well-designed for
© 2020 Copyright for this paper by its authors. Use permitted under Creative CPWrEooUrckReshdoinpgs IhStpN:/c1e6u1r3-w-0s.o7r3g CCoEmUmoRns WLiceonrsekAsthtriobuptioPnr4o.0cIneteerdnaitniognasl ((CCC EBYU4R.0)-.WS.org)
Some existing studies have focused on the correlation between learning and users’ searching progress. Researchers such as Vakkari [3] and Wildemuth [4] have explored the role of information searching in learning, and the factors or concepts related to searching as learning. Freund et al. [5] found that reading could be the core component of users’ searching progress, and they studied the text on the impact of learning outcomes.
Extensive studies have also been conducted on searching systems or techniques in supporting learning. Stange et al. [6] investigated users’ searching progress and found that individuals could integrate their knowledge gains into the searching process through concept maps, and understand the relationship between learning, sense making, and information seeking. Egusa et al. [7] revealed that the visualization tools in search engines, such as concept maps, could be used to evaluate evolvement of users knowledge structure and search behaviours. Jansen et al. [8] classified searching as learning tasks of users to verify whether there were specific factors in the learning process. Moreover, they found that web searchers relied more on their knowledge and information needs, while the searching was used mainly to check facts.
There were also studies focusing on how the performance of searching as learning can be improved by assistance of new models, systems, and other methods. Saito et al. [9] designed a searching platform that supports users thinking activities. The platform visualized learners’ searching process and promote their thinking by comparing learners’ searching as learning processes to those of other searchers. Bah and Carterette [10] created a system to support continuous searching as learning. The system firstly created typical pseudo-documents and then sorted information from retrieval results according to how closely it matched the typical document.
Existing studies have shown that learning is embedded in the searching processes. However, these works are focused more on developing dedicated systems and single-user scenarios, and more studies are needed that focus on searching as learning within search engines and more complex scenarios such as collaborative web search.
One of the first works regarding developing an interface for collaborative learning was published in 1996 by Michael B. Twidale and David M. Nichols in their work “Interfaces to support collaboration in information retrieval” [11]. The key idea was to develop an interface, which allows to collect the user’s queries and their results, and after that to visualize the search process.
A more recent solution is SearchX [12], which is based on the search engine Pineapple Search1. SearchX2 is a search system, which includes a collaborative search interface. People can collaborate in groups during the searching process by using diferent widgets, such as shared query history with the groupmates, bookmarks of useful information and sites which can be seen and used by other people in the group.
Systems such as popHistory [13] and Warcbase [14, 15] save users’ visit data, based on which they can extract and display the most visited websites to users. History Viewer [16] tracks processes of exploratory search and presents users with interaction data to enable them to revisit the steps that led to particular insights.
Information re-finding tools such as SearchBar [17]
provide a hierarchical history of recent search topics,
queries, results and users’ notes to help users quickly
re-find the information they have searched. The
system Personal Web Library helps users to understand
their web browsing patterns, identify their topics of
interest and retrieve previously visited webpages more
easily [18]. Some other tools, such as SIS (Stuf I’ve
Seen) [
In collaborative search systems such as Coagmento [24] and SearchTogether [25], visualization of search history usually involves multiple users’ search logs, including their search queries, and bookmarks. Interfaces of this kind display search histories separately according to data types or categories and support notepad functions which allow group members to share an experience.
By contrast to the systems described above, LogCanvas v2 provides a visualization of the search sessions based on a timeline and tags the activities of users. Our aim is to help users to reconstruct easily their
searching process and re-find an information which they searched for.
In this section, we compare history-related functions in three searching platforms: Google Search, Pocket, and Learnweb3. Learnweb is a web platform, developed for searching, collecting and sharing educational resources [26, 27]. In order to understand what might be missing in the current version of Learnweb, we investigated which functionalities are ofered to the users from their search histories. As shown in Table 1, we compared the history-related functions of the three platforms from diferent perspectives: (i) history retrieving (i.e. searching scope, searching method), (ii) bookmarking (i.e. bookmarking method, operations on bookmarked resources, and bookmarking organizing method), and (iii) support for learning.
History retrieving Being able to review queries and search results quickly is considered as a highly useful feature of a history interface. In all the three reviewed interfaces, users can review their search histories (i.e. queries, search results and bookmarked resources). However, there are still diferences as regards history retrieving scope and retrieving methods. 1. Searching scope. Since Learnweb was developed to be a collaborative searching and archiving sys3https://learnweb.l3s.uni-hannover.de/ tem to support users independent or collaborative search, its history interface was designed to support varied retrieving scopes including: private resources, wherein a user’s individual searching records can be found; topical groups, wherein queries and search results of members in the user’s collaboration group can be found; archived documents, wherein the search results archived by the user can be found. On the contrary, Pocket and Google Search support retrieval within users’ individual searching records. 2. Searching method. In the Learnweb platform and in Google Search users can do a full-text search to retrieve histories. Pocket only provides keyword search and search by URL (exact matching).
Bookmarking By applying the “bookmark” function during searching, users can mark important search results for quickly information re-finding or reviewing in the future. In the three analysed platforms, bookmarking methods and supporting operations on bookmarked resources are diferent.
Search, users can directly bookmark search results through a pop-up guideline within the platform. When using Pocket to store webpages, a Pocket plugin should be added to the search engine. 2. Operation on bookmarked/archived resources.
Using the Learnweb platform and Pocket users can edit metadata of archived resources including ratings, tags and comments. Besides this, in
Pocket users can select and highlight the text of and project-based learning [30], short term or long term, a resource and review all the highlights. In the can be achieved. In order to achieve TBL, students Chrome browser, users can only change the file should work in small groups, be accountable for the name and the file location. work they are performing, and receive feedback [31]. 3. Organizing bookmarked/archived resources. A In order to achieve the maximum impact in learning, possibility of organizing saved resources is pre- the groups should work on diferent parts of the same sent in all three interfaces. In Learnweb, a user problem, which demonstrates a useful concept and recan organize the resources by topical groups, port simultaneously. Focused on these goals we defolders and tags; in Pocket by tags and in Chrome veloped a learning scenario to test our platform (Secby folders. tion 5.1).
cilitate distance learning. By using web resources, stu- 4. The Design of LogCanvas v2.0 dents can learn asynchronously from any place with an internet connection. According to our analysis, Having a better knowledge of existing solutions, we Learnweb is outstanding in terms of supporting learn- started to design a new search history interface to be ing during searching. It provides learning-related func- integrated into the Learnweb platform. tions that allow users to work in topical groups, join specific courses and perform data analysis of their work 4.1. Overview or their groupmates [28]. On the other hand, we noticed that Pocket, as a history manager, supports annotating for articles or videos in bookmarked webpages where users can directly add their comments or personal ideas to the saved resources. This kind of interaction between users and search history is shown to be efective in supporting learning during the information searching/re-finding process [29]. Thus, we explored the possibility of adding a similar function to Learnweb (as described in Section 4).
Moreover, in Learnweb team-based learning (TBL)
recover the search history from the Learnweb platform. The interface was inspired by diferent webbrowsers, such as Google Chrome, Firefox and Safari.
The latest version of the interface focuses more on the users’ queries and has several new features in comparison to the previous versions.
In order to record the searching process, first of all, users have to perform some search queries in Learnweb.
After registration and login into the platform, the user is redirected to the main page and can perform a search 4.2. Collaborative Work with using the navigation panel, which is also located on Annotation Features each page of the platform, as it is visible in Figure 1 (A).
Each internal webpage also has a breadcrumb naviga- During a preliminary evaluation of our prototype, we tion (e.g. “Search history”) as shown in Figure 1 (B). noticed that a connection between the search history
The resulting search history page is divided into two results and the original searching process was missparts: a search sessions list with all logged queries, ing. In particular, while revisiting the search history where in Figure 1 (C) the query “shaman...ingredients” results and the resources stored in the personal folders, is selected to check the corresponding search history; it was dificult for the user to reconstruct the searching and a list of the archived web results (i.e. snippets) as context and the links to the original webpages; for exdisplayed in Figure 1 (D). To recover the search his- ample to remember why a specific resource had been tory and snippets, a user has to choose a query from selected to be stored instead of others. the sessions list panel. This panel includes three ele- For this reason, we decided to introduce webpage ments: a button to switch between the personal search annotations in order to provide the missing informahistory and the history generated by other members of tion. Among other annotation tools we choose Hya common group, a search bar that allows quick filter pothes.is4 and integrated their client into Learnweb through all users’ sessions and queries, and a timeline ecosystem. It allows taking notes directly on online of all search sessions by descending order. webpages or in PDF-documents, and displays an over
By clicking on one of the queries from the session view of all annotation activities, as it shown on Figlist, a list of snippets of archived search results will ure 2, where all the highlighted and annotated text be opened on the right side of the window, wherein were taking on the webpage of “Shaman Energy Drink”. all visited snippets are highlighted for a quick review. All annotated parts of the webpage are shown in the The visited snippet “Shaman Energy Drink” is high- right panel (Fig. 2 (B)). By clicking on one of the annolighted as an example in Figure 1 (D). A filter allows to tations, on the webpage will be highlighted the chosen choose which snippets to visualize: only the snippets part of the text (Fig. 2 (A)), and vice versa. More specifthat were visited during the searching process, or to ically, in Figure 2 (A) the annotated website where the display all the snippets as they were shown in the orig- content “SHAMAN...long-lasting..” is highlighted; and inal rank on the web results page. The clicked snippets in Figure 2 (B) the annotating panel containing all comare highlighted for easier recovering of the searching ments of team members is displayed, where in our exprocess (marked with green and has the hand icon). ample the user “tania_tol” added an annotation “Ingre- and the learning components of learning scenarios. dients” to the previous highlighted content. Title: Which energy drinks are poor for health? Target group: High school students (15-18 years old) 4.3. Learning Assumption in Chemistry class, working in teams of 2-3 people. Estimated duration: 3 hours.
For the evaluation of the LogCanvas v2 search his- Learning environment: Online. tory interface, we formulate the following learning as- Learning outcomes: To research and gather inforsumption, which we plan to evaluate with a specific mation from online search; to know about chemical learning scenario. We argue that an annotation sys- components that commonly exist in energy drinks and tem in combination with the search history can their efects on health; to learn which energy drinks be a useful tool for users in group projects, and contain (larger) doses of unhealthy components; to encan improve the teaching and learning experience. hance learning through student collaboration and new In previous studies [32, 33], we find remarks for ben- forms of assessments; to work and collaborate in a efits of using annotations, such as highlighted text, team; to organize ideas and data, and present them in comments, and tags, as a useful and easy way to achieve front of their peers; to organize and present a final recollective work in an online domain. Annotations on a port. search base can reveal the insights of group activity in Learning scenario: Students are asked to form groups collaborative work. Hence, they help teachers and stu- of 2 to 3 people in the chemistry class. The teacher dents track the thought and learning process by iden- presents the project definition and explains the two tifying the search results over time, and at the same parts of the project. In the first part, students should time document the references which were used to find do research in order to identify the most common unthe students’ answers. This is profitable for students healthy ingredients in the energy drinks. Second, they because based on the search results of their classmates, will pick an ingredient - after discussing with the teathey decide to revisit a webpage or not. Documenting cher, so each team picks a diferent one - and search the search process with annotations can help students, about the energy drinks which include this ingredient; also in motivating their choices while preparing the or if all energy drinks contain it, then pick the ones ifnal course presentation. Further, it is efective for with higher concentration. The final results of each teachers as they can view the students’ activity and team will be reported in a brief report and presented trace their resources. Therefore, we hypothesize that in front of the class. joint work projects can be achieved in the Learnweb Future perspectives: This scenario can be further explatform successfully. tended as a multidisciplinary project with the sports science class. Additionally, it could interest university 5. Experimental Scenario lseimctiularersleraerlantiendg tsocehneaarltioh,adsie(ptaarntdofs)poartpsrotojecpte-rbfaosremd evaluation.
Evaluation: The students present their results in a report and prepare a presentation.
signed to test the learning assumption on the Learnweb search engine (Section 5.1), as well the user experience while completing the learning scenario in Learnweb (Section 5.2).
The Learnweb platform ofers the ability to perform sites containing useful information about energy drinks. a wide variety of learning scenarios, based on the re- In a preliminary experiment, students needed about quirements and the needs of each course. Team-based, 10 queries and visited about 20 websites to find all reproject-based, and distance learning scenarios, like the quired answers. A similar search process is displayed one described here, can successfully be executed. in Figure 1. These answers can be highlighted/annota
The design of this learning scenario was based on ted in the new Learnweb platform, and all highlighted the essential elements for TBL. Its development is in- texts, annotations and websites are added to a spelfuenced by a template [34] and examples for collabo- cial annotations file in the students work-space. This rative learning5 that analytically present the structure helps students document their work and ofers them the ability to prove their process to other team mem5http://colab.eun.org/learning-scenarios bers; by the timestamp and users name on each anno
In this paper, we introduce LogCanvas v2 - a new and extended design of the search history visualization in our Learnweb educational platform. The new interface documents the complete search and learning process of students in a distance learning, team-based and project-based learning scenario. All visited websites are documented, relevant sections of the websites which provide answers to the student are highlighted, annotated and discussed by the students. This makes the writing of the final report much easier for the students and makes sure all references are included in this report. tation. The latter promotes their individual accountability and aids the accountability of each member for contributing to the team based on the TBL [31]. Furthermore, it creates a proof of work for the work carried by each member to the teacher.
Additionally, it helps other team members realise the results of their teammates tag, share, highlight or comment them, as it is displayed in Figure 2. In this way, the team can collaborate and each team member can critically comment others observations and findings. Further, it ofers the ability for quick and precise feedback from the teachers, for the intermediate outcomes.
Finally, the annotations file is helpful for the final evaluation. It can help students identify quickly the important webpages and the specific parts of useful information, which they have annotated during their work on the project. Moreover, it easily links to all used resources, letting no information untraceable. Furthermore, it displays each member contribution to the team’s output; by checking the annotations file; and the collaboration’s between the team; by analysing the comments, tags and highlights to other team members content.