Figure 1 - uploaded by Fabio Gasparetti
Content may be subject to copyright.
Implicit Feedback techniques extract information related to the current user needs
Source publication
Browsing activities are an important source of information to build profiles of the user interests and personalize the human- computer interaction during information seeking tasks. Vis- ited pages are easily collectible, e.g., from browsers' histo- ries and toolbars, or desktop search tools, and they often con- tain documents related to the current...
Context in source publication
Context 1
... bur- den on the users is still high and the benefits are not always clear compared with other approaches [11]. Implicit Feedback techniques [7] passively monitor the user behavior gathering usage data to build a profile of the user needs (see Fig.1). Users do not have to explicitly indicate which documents are relevant. ...
Similar publications
Websites using content as their main asset need to ensure a steady revenue flow from their digital projects. Along with the subscription models, this is mainly achieved by the efficient delivery of advertisements (ads) to their readers. Website users stand as an opposition to this initiative, protecting their privacy and trying to improve their use...
Citations
... Many information systems already utilise smart features such as recommender systems to curate and present data to users (Atvisor, 2021), aiming to make the most relevant information more quickly accessible. Previous work has shown opportunities to tailor and improve user experiences by building user profiles through visited pages and toolbars (Gasparetti & Micarelli, 2007) and predicting search behaviours through a browser plug-in (Mitsui et al., 2017). Yet the algorithms subsequently used to index, retrieve, and rank data are often driven by underlying data that are themselves biased (Gao & Shah, 2021). ...
... Researchers have experimented with clustering approaches with and without human intervention. While unsupervised learning algorithms can automatically cluster similar user preferences to facilitate interface design (Salminen et al., 2020;Guan et al., 2016;Gasparetti and Micarelli, 2007), frequently, collaborations between experimenters, experts, and unsupervised algorithms are necessary when clustering based on unstructured information, such as audio and visual data (Chuang et al., 2012). On the other end of the spectrum, Pruitt and Grudin (2003), for instance, used an experimenter-driven approach that clustered qualitative and quantitative evidence to identify groups of user preferences. ...
Digital reading applications give readers the ability to customize fonts, sizes, and spacings, all of which have been shown to improve the reading experience for readers from different demographics. However, tweaking these text features can be challenging, especially given their interactions on the final look and feel of the text. Our solution is to offer readers preset combinations of font, character, word and line spacing, which we bundle together into reading themes. To arrive at a recommended set of reading themes, we present our THERIF pipeline, which combines crowdsourced text adjustments, ML-driven clustering of text formats, and design sessions. We show that after four iterations of our pipeline, we converge on a set of three COR themes (Compact, Open, and Relaxed) that meet diverse readers' preferences, when evaluating the reading speeds, comprehension scores, and preferences of hundreds of readers with and without dyslexia, using crowdsourced experiments.
... The authors of Gasparetti and Micarelli (2007) propose to solve this problem by first starting of the leaves of the DOM-based representation of a page, collects each node v whose tag is in T ags, which contains significant amount of text. The pages is therefore split in units whose boundaries are arranged by HTML tags and the text is retrieved by the deepest units. ...
Browsing sessions are rich in elements useful to build profiles of user interests, but at the same time HTML pages include noisy data such as advertisements, navigation menus and privacy notes. Moreover, some pages cover several different topics making it difficult to identify the most relevant to the user. For these reasons, they are often ignored by personalized search and recommender systems. We propose a novel approach for recognizing valuable text descriptions of current user information needs—namely cues—based on the data mined from browsing interactions over the web. The approach combines page clustering techniques based on Document Object Model-based representations for acquiring evidence about relevant correlations between text contents. This evidence is exploited for better filtering out irrelevant information and facilitating the construction of interest profiles. A comparative framework proves the accuracy of the extracted cues in the personalize search task, where results are re-ranked according to the last browsed resources.
... In [9,10], the authors propose a preliminary attempt for extracting cues of user information needs based only on the user's most recent activity. In that model, the browsing activity is exploited in order to identify a set of words that characterize the current needs of the user. ...
The time spent using a web browser on a wide variety of tasks such as research activities, shopping or planning holidays is relevant. Web pages visited by users contain important hints about their interests, but empirical evaluations show that almost 40-50% of the elements of the web pages can be considered irrelevant w.r.t. the user interests driving the browsing activity. Moreover, pages might cover several di�erent topics. For these reasons they are often ignored in personalised approaches. We propose a novel approach for selectively collecting text information based on any implicit signal that naturally exists through web browsing interactions. Our approach consists of three steps: (1) definition of a DOM-based representation of visited pages, (2) clustering of pages according with a tree edit distance measure and (3) exploiting the acquired evidence about the user behaviour to better �lettering out irrelevant information and identify relevant text related to the current needs. A comparative evaluation shows the e�ectiveness of the proposed approach in retrieving additional web resources related to what the user is currently browsing
... In [9,10], the authors propose a preliminary attempt for extracting cues of user information needs based only on the user's most recent activity. In that model, the browsing activity is exploited in order to identify a set of words that characterize the current needs of the user. ...
The time spent using a web browser on a wide variety of tasks such as research activities, shopping or planning holidays is relevant. Web pages visited by users contain important hints about their interests, but empirical evaluations show that almost 40-50% of the elements of the web pages can be considered irrelevant w.r.t. the user interests driving the browsing activity. Moreover, pages might cover several different topics. For these reasons they are often ignored in personalized approaches. We propose a novel approach for selectively collecting text information based on any implicit signal that naturally exists through web browsing interactions. Our approach consists of three steps: (1) definition of a DOM-based representation of visited pages, (2) clustering of pages according with a tree edit distance measure and (3) exploiting the acquired evidence about the user behaviour to better filtering out irrelevant information and identify relevant text related to the current needs. A comparative evaluation shows the effectiveness of the proposed approach in retrieving additional web resources related to what the user is currently browsing.
... In [5], the authors propose a preliminary attempt for extracting cues of user information needs based only on the user's most recent activity. In that model, the browsing activity is exploited in order to identify a set of words that characterize the current needs of the user. ...
Browsing sessions are rich in elements useful to build profiles of user interests, but at the same time HTML pages include noise data, such as ads and navigation menus. Moreover, pages might cover several different topics. For these reasons they are often ignored in personalized approaches. We propose a novel approach for implicitly recognizing valuable text descriptions of current user needs based on the implicit feedback revealed through web browsing interactions.
... Both explicit and implicit user profiling techniques have been investigated in an effort to improve the search results from the web. The implicit profiling method is concerned with the observation of the behaviour of the users and capturing their search history [8]. One main advantage of this approach is that it is less intrusive and does not require user involvement. ...
This paper is concerned with the issue of low quality of web search results and ways of improving it. Information sources are exponentially increasing on the web and are making it increasingly difficult to get relevant information. The quality of web search results depends on two factors: the information needs of the users and the searching techniques. In order to improve the quality of search results it is necessary to have information about user requirements. This paper attempts to address the issue of low quality of web search results through the user profiling technique, where the profiles can be explicitly specified or implicitly generated. The proposed approach in this paper is based on explicit user profiles. The implemented system allows users to create, save, remove and update profiles based on their changing information needs and preferences. Preliminary experimental results indicate that the system produces more refined results based on explicit user profiles when compared to traditional search methods.
... Through our zoning concept, we can exclude the unnecessary content like possibly navigation menu, advertisement, contact information or copyrights. Thus it is similar to the aim of Gasparetti et al. [4], who analyze parts of the HTML of visited pages to identify user needs. However they have no possibility to customize a strategy and a specific semantics for different business goals. ...
In this paper, we report our experience in the implementation of a module for creating user profiles of Web visitors by using Zones, Weights and Actions. The module is part of Gugubarra 2.0, a tool for better understanding and management of communities of registered Web visitors, currently being developed by the database group at the Computer Science Institute of the Goethe University in Frankfurt. In addition we introduce a semantic concept for our approach.
... The text comparison is based on the IR similarity function used in the TextTiling algorithm. 50 For a deeper analysis of the cue-extraction technique, see Ref. 51. ...
Personalized retrieval of documents is a research field that has been gaining interest, since it is a possible solution to the information overload problem. The ability to adapt the retrieval process to the current user needs increases the accuracy and reduces the time users spend to formulate and sift through result lists. In this chapter we show two instances of user modeling. One is based on the human memory theory named Search of Associative Memory, and a further approach based on the Hyperspace Analogue to Language model. We prove how by implicit feedback techniques we are able to unobtrusively recognize user needs and monitor the user working context. This is important to provide personalization during traditional information retrieval and for recommender system development. We discuss an evaluation comparing the two cognitive approaches, their similarities and drawbacks. An extended analysis reveals interesting evidence about the good performance of SAM-based user modeling, but it also proves how HAL-based models evaluated in the Web browsing context shows slightly higher degree of precision.
