Click-through rate (CTR) is a key signal of relevance for search engine results, both organic and sponsored. CTR of a result has two core components: (a) the probability of examination of a result by a user, and (b) the perceived relevance of the result given that it has been examined by the user. There has been considerable work on user browsing models, to model and analyze both the examination and the relevance components of CTR. In this paper, we propose a novel formulation: a micro-browsing model for how users read result snippets. The snippet text of a result often plays a critical role in the perceived relevance of the result. We study how particular words within a line of snippet can influence user behavior. We validate this new micro-browsing user model by considering the problem of predicting which snippet will yield higher CTR, and show that classification accuracy is dramatically higher with our micro-browsing user model. The key insight in this paper is that varying relatively few words within a snippet, and even their location within a snippet, can have a significant influence on the clickthrough of a snippet.

People with visual disabilities, especially those who are blind, have digital content narrated to them by text-to-speech (TTS) engines (e.g., with the help of screen readers). Naively narrating web pages, particularly the ones consisting of several diverse pieces (e.g., news summaries, opinion pieces, taxonomy, ads), with TTS engines without organizing them into thematic segments will make it very difficult for the blind user to mentally separate out and comprehend the essential elements in a segment, and the effort to do so can cause significant cognitive stress. One can alleviate this difficulty by segmenting web pages into thematic pieces and then narrating each of them separately. Extant segmentation methods typically segment web pages using visual and structural cues. The use of such cues without taking into account the semantics of the content, tends to produce "impure" segments containing extraneous material interspersed with the essential elements. In this paper, we describe a new technique for identifying thematic segments by tightly coupling visual, structural, and linguistic features present in the content. A notable aspect of the technique is that it produces segments with very little irrelevant content. Another interesting aspect is that the clutter-free main content of a web page, that is produced by the Readability tool and the "Reader" feature of the Safari browser, emerges as a special case of the thematic segments created by our technique. We provide experimental evidence of the effectiveness of our technique in reducing clutter. We also describe a user study with 23 blind subjects of its impact on web accessibility.

In our information-driven web-based society, we are all gradually falling ""victims"" to information overload [5].However, while sighted people are finding ways to sift through information faster, Internet users who are blind are experiencing an even greater information overload. These people access computers and Internet using screen-reader software, which reads the information on a computer screen sequentially using computer-generated speech. While sighted people can learn how to quickly glance over the headlines and news articles online to get the gist of information, people who are blind have to use keyboard shortcuts to listen through the content narrated by a serial audio interface. This interface does not give them an opportunity to know what content to skip and what to listen to. So, they either listen to all of the content or listen to the first part of each sentence or paragraph before they skip to the next one. In this paper, we propose an automated approach to facilitate non-visual skimming of web pages. We describe the underlying algorithm, outline a non-visual skimming interface, and report on the results of automated experiments, as well as on our user study with 23 screen-reader users. The results of the experiments suggest that we have been moderately successful in designing a viable algorithm for automatic summarization that could be used for non-visual skimming. In our user studies, we confirmed that people who are blind could read and search through online articles faster and were able to understand and remember most of what they have read with our skimming system. Finally, all 23 participants expressed genuine interest in using non-visual skimming in the future.

Abstract Probabilistic Logic Programming (PLP), exemplified by Sato and Kameya's PRISM, Poole's ICL, Raedt et al.'s ProbLog and Vennekens et al.'s LPAD, is aimed at combining statistical and logical knowledge representation and inference. However, the inference techniques used in these works rely on enumerating sets of explanations for a query answer. Consequently, these languages permit very limited use of random variables with continuous distributions. In this paper, we present a symbolic inference procedure that uses constraints and represents sets of explanations without enumeration. This permits us to reason over PLPs with Gaussian or Gamma-distributed random variables (in addition to discrete-valued random variables) and linear equality constraints over reals. We develop the inference procedure in the context of PRISM; however the procedure's core ideas can be easily applied to other PLP languages as well. An interesting aspect of our inference procedure is that PRISM's query evaluation process becomes a special case in the absence of any continuous random variables in the program. The symbolic inference procedure enables us to reason over complex probabilistic models such as Kalman filters and a large subclass of Hybrid Bayesian networks that were hitherto not possible in PLP frameworks.

People who are blind use screen readers for browsing web pages. Since screen readers read out content serially, a naive readout tends to mix irrelevant and relevant content thereby disrupting the coherency of the material being read out and confusing the listener. To address this problem we can partition web pages into coherent segments and narrate each such piece separately. Extant methods to do segmentation use visual and structural cues without taking the semantics into account and consequently create segments containing irrelevant material. In this paper, we describe a new technique for creating coherent segments by tightly coupling visual, structural, and linguistic features present in the content. A notable aspect of the technique is that it produces segments with little irrelevant content. Preliminary experiments indicate that the technique is effective in creating highly coherent segments and the experiences of an early adopter who is blind suggest that it enriches the overall browsing experience.

This demonstration will propose a touch-based directional navigation technique, on touch interface (e.g., iPhone, Macbook) for people with visual disabilities especially blind individuals. Such interfaces coupled with TTS (text-to-speech) systems open up intriguing possibilities for browsing and skimming web content with ease and speed. Apple's seminal VoiceOver system for iOS is an exemplar of bringing touch-based web navigation to blind people. There are two major shortcomings: "fat finger" and "finger-fatigue" problems, which have been addressed in this paper with two proposed approaches. A preliminary user evaluation of the system incorporating these ideas suggests that they can be effective in practice.

Advances in web technology have considerably widened the Web accessibility divide between sighted and blind users. This divide is especially acute when conducting online transactions, e.g., shopping, paying bills, making travel plans, etc. Such transactions span multiple web pages and require that users find clickable objects (e.g., “add-to-cart” button) which are essential for transaction progress. While this is fast and straightforward for sighted users, locating the clickable objects causes considerable strain for blind individuals using screen-reading technology. Screen readers force users to listen to irrelevant information sequentially and provide no interface for identifying relevant clickable objects. This paper addresses the problem of making clickable objects readily accessible, which can substantially reduce the information overload that is otherwise experienced by blind users. A static knowledge base of keywords constructed from the captions of clickable objects does not provide enough learning capability for identifying clickable objects which do not have any captions (e.g., image buttons without alternative text). In this paper, we present an Information Retrieval based technique that uses the context of transaction-centric objects (e.g., “add-to-cart” and “checkout” buttons) to identify and classify them even when their captions are missing. In addition, the technique utilizes a reinforcement mechanism based on user feedback to accommodate previously unseen captions of objects as well as new categories of objects. We provide user study and experimental evidence of the effectiveness of our algorithm.

An important aspect of making the Web accessible to blind users is ensuring that all important web page elements such as links, clickable buttons, and form fields have explicitly assigned labels. Properly labeled content is then correctly read out by screen readers, a dominant assistive technology used by blind users. In particular, improperly labeled form fields can critically impede online transactions such as shopping, paying bills, etc. with screen readers. Very often labels are not associated with form fields or are missing altogether, making form filling a challenge for blind users. Algorithms for associating a form element with one of several candidate labels in its vicinity must cope with the variability of the element's features including label's location relative to the element, distance to the element, etc. Probabilistic models provide a natural machinery to reason with such uncertainties. In this paper we present a Finite Mixture Model (FMM) formulation of the label association problem. The variability of feature values are captured in the FMM by a mixture of random variables that are drawn from parameterized distributions. Then, the most likely label to be paired with a form element is computed by maximizing the log-likelihood of the feature data using the Expectation-Maximization algorithm. We also adapt the FMM approach for two related problems: assigning labels (from an external Knowledge Base) to form elements that have no candidate labels in their vicinity and for quickly identifying clickable elements such as add-to-cart, checkout, etc., used in online transactions even when these elements do not have textual captions (e.g., image buttons w/o alternative text). We provide a quantitative evaluation of our techniques, as well as a user study with two blind subjects who used an aural web browser implementing our approach.

This demo will present HearSay, a multi-modal non-visual web browser, which aims to bridge the growing Web Accessibility divide between individuals with visual impairments and their sighted counterparts, and to facilitate full participation of blind individuals in the growing Web-based society.

In this paper, we propose two cooperative ensemble learning algorithms, i.e., NegBagg and NegBoost, for designing neural network (NN) ensembles. The proposed algorithms incrementally train different individual NNs in an ensemble using the negative correlation learning algorithm. Bagging and boosting algorithms are used in NegBagg and NegBoost, respectively, to create different training sets for different NNs in the ensemble. The idea behind using negative correlation learning in conjunction with the bagging/boosting algorithm is to facilitate interaction and cooperation among NNs during their training. Both NegBagg and NegBoost use a constructive approach to automatically determine the number of hidden neurons for NNs. NegBoost also uses the constructive approach to automatically determine the number of NNs for the ensemble. The two algorithms have been tested on a number of benchmark problems in machine learning and NNs, including Australian credit card assessment, breast cancer, diabetes, glass, heart disease, letter recognition, satellite, soybean, and waveform problems. The experimental results show that NegBagg and NegBoost require a small number of training epochs to produce compact NN ensembles with good generalization.