Recent advances in readability assessment have lead to the introduction of multilingual strategies that can predict the reading-level of a text regardless of its language. These strategies, however, tend to be limited to just operating in different languages rather than taking any explicit advantage of the multilingual corpora they utilize. In this manuscript, we discuss the results of the in-depth empirical analysis we conducted to assess the language transfer capabilities of four different strategies for readability assessment with increasing multilingual power. Results showcase that transfer learning is a valid option for improving the performance of readability assessment, particularly in the case of typologically-similar languages and when training corpora availability is limited.

@en

Most research efforts related to automatic readability assessment focus on the design of strategies that apply to a specific language. These state-of-the-art strategies are highly dependent on linguistic features that best suit the language for which they were intended, constraining their adaptability and making it difficult to determine whether they would remain effective if they were applied to estimate the level of difficulty of texts in other languages. In this article, we present the results of a study designed to determine the feasibility of a cross-lingual readability assessment strategy. For doing so, we first analyzed the most common features used for readability assessment and determined their influence on the readability prediction process of 6 different languages: English, Spanish, Basque, Italian, French, and Catalan. In addition, we developed a cross-lingual readability assessment strategy that serves as a means to empirically explore the potential advantages of employing a single strategy (and set of features) for readability assessment in different languages, including interlanguage prediction agreement and prediction accuracy improvement for low-resource languages.

@en

The alignment of word embedding spaces in different languages into a common crosslingual space has recently been in vogue. Strategies that do so compute pairwise alignments and then map multiple languages to a single pivot language (most often English). These strategies, however, are biased towards the choice of the pivot language, given that language proximity and the linguistic characteristics of the target language can strongly impact the resultant crosslingual space in detriment of topologically distant languages. We present a strategy that eliminates the need for a pivot language by learning the mappings across languages in a hierarchical way. Experiments demonstrate that our strategy significantly improves vocabulary induction scores in all existing benchmarks, as well as in a new non-English–centered benchmark we built, which we make publicly available.

@en

To capitalize on the benefits associated with word embeddings, researchers working with data from domains such as medicine, sentiment analysis, or finance, have dedicated efforts to either taking advantage of popular, general-purpose embedding-learning strategies, such as Word2Vec, or developing new ones that explicitly consider domain knowledge in order to generate new domain-specific embeddings. In this manuscript, we instead propose a mixed strategy to generate enriched embeddings specifically designed for the educational domain. We do so by leveraging FastText embeddings pre-trained using Wikipedia, in addition to established educational standards that serve as structured knowledge sources to identify terms, topics, and subjects for each school grade. The results of an initial empirical analysis reveal that the proposed embedding-learning strategy, which infuses limited structured knowledge currently available for education into pre-trained embeddings, can better capture relationships and proximity among education-related terminology. Further, these results demonstrate the advantages of using domain-specific embeddings over general-purpose counterparts for capturing information that pertains to the educational area, along with potential applicability implications when it comes to text processing and analysis for K-12 curriculum-related tasks.

@en

We present a multiattentive recurrent neural network architecture for automatic multilingual readability assessment. This architecture considers raw words as its main input, but internally captures text structure and informs its word attention process using other syntax-and morphology-related datapoints, known to be of great importance to readability. This is achieved by a multiattentive strategy that allows the neural network to focus on specific parts of a text for predicting its reading level. We conducted an exhaustive evaluation using data sets targeting multiple languages and prediction task types, to compare the proposed model with traditional, state-of-the-art, and other neural network strategies.

@en

Popular search engines are usually tuned to satisfy the information needs of a general audience. As a result, non-traditional, yet active groups of users, such as children, experience challenges composing queries that can lead them to the retrieval of adequate results. To aid young users in formulating keyword queries that can facilitate their information-seeking process, we introduce ReQuIK, a multi-perspective query suggestion system for children. ReQuIK informs its suggestion process by applying (i) a strategy based on search intent to capture the purpose of a query, (ii) a ranking strategy based on a wide and deep neural network that considers both raw text and traits commonly associated with kid-related queries, (iii) a filtering strategy based on the readability levels of documents potentially retrieved by a query to favor suggestions that trigger the retrieval of documents matching children’s reading skills, and (iv) a content-similarity strategy to ensure diversity among suggestions. For assessing the quality of the system, we conducted initial offline and online experiments based on 591 queries written by 97 children, ages 6 to 13. The results of this assessment verified the correctness of ReQuIK’s recommendation strategy, the fact that it provides suggestions that appeal to children and ReQuIK’s ability to recommend queries that lead to the retrieval of materials with readability levels that correlate with children’s reading skills.

@en

In the research literature, evaluations of recommender system effectiveness typically report results over a given data set, providing an aggregate measure of effectiveness over each instance (e.g. user) in the data set. Recent advances in information retrieval evaluation, however, demonstrate the importance of considering the distribution of effectiveness across diverse groups of varying sizes. For example, do users of different ages or genders obtain similar utility from the system, particularly if their group is a relatively small subset of the user base? We apply this consideration to recommender systems, using offline evaluation and a utility-based metric of recommendation effectiveness to explore whether different user demographic groups experience similar recommendation accuracy. We find demographic differences in measured recommender effectiveness across two data sets containing different types of feedback in different domains; these differences sometimes, but not always, correlate with the size of the user group in question. Demographic effects also have a complex—and likely detrimental—interaction with popularity bias, a known deficiency of recommender evaluation. These results demonstrate the need for recommender system evaluation protocols that explicitly quantify the degree to which the system is meeting the information needs of all its users, as well as the need for researchers and operators to move beyond naïve evaluations that favor the needs of larger subsets of the user population while ignoring smaller subsets.@en

Curation is the act of selecting, organizing, and presenting content. Some applications emulate this process by turning users into curators, while others use recommenders to select items, seldom achieving the focus or selectivity of human curators. We bridge this gap with a recommendation strategy that more closely mimics the objectives of human curators. We consider multiple data sources to enhance the recommendation process, as well as the quality and diversity of the provided suggestions. Further, we pair each suggestion with an explanation that showcases why a book was recommended with the aim of easing the decision making process for the user. Empirical studies using Social Book Search data demonstrate the effectiveness of the proposed methodology.

@en

The increasing complexity and ubiquity in user connectivity, computing environments, information content, and software, mobile, and web applications transfers the responsibility of privacy management to the individuals. Hence, making it extremely difficult for users to maintain the intelligent and targeted level of privacy protection that they need and desire, while simultaneously maintaining their ability to optimally function. Thus, there is a critical need to develop intelligent, automated, and adaptable privacy management systems that can assist users in managing and protecting their sensitive data in the increasingly complex situations and environments that they find themselves in. This work is a first step in exploring the development of such a system, specifically how user personality traits and other characteristics can be used to help automate determination of user sharing preferences for a variety of user data and situations. The Big-Five personality traits of openness, conscientiousness, extroversion, agreeableness, and neuroticism are examined and used as inputs into several popular machine learning algorithms in order to assess their ability to elicit and predict user privacy preferences. Our results show that the Big-Five personality traits can be used to significantly improve the prediction of user privacy preferences in a number of contexts and situations, and so using machine learning approaches to automate the setting of user privacy preferences has the potential to greatly reduce the burden on users while simultaneously improving the accuracy of their privacy preferences and security.

@en

Information discovery tasks using online search tools are performed on a regular basis by school-age children. However, these tools are not necessarily designed to both explicitly facilitate the retrieval of resources these young users can comprehend and aid low-literacy searchers. This is of particular concern for educational environments, as there is an inherent expectation that these tools facilitate effective learning. In this manuscript we present an initial assessment conducted over (1) children-oriented search tools based on queries generated by K-9 students, analyzing features such as readability and adequacy of retrieved results, and (2) tools used by teachers in their classrooms, analyzing their main purpose and target audience’s age range. Among the examined tools, we include YouUnderstood.Me, an enhanced search environment, which is the result of our ongoing efforts on the development of a search environment tailored to 5-15 year-olds that can foster learning through the retrieval of materials that not only satisfy the information needs of these users but also match their reading abilities. The results of these studies highlight the fact that search results presented to children have average reading levels that do not match the target audience. In addition, tools oriented to teachers do not go beyond showing the progress of their students, and seldomly provide a simple way of retrieving class contents that fit current needs of students. These facts further showcase the need for developing a dual environment oriented to both teachers and students.

@en

We present an initial examination on the impact text complexity has when incorporated into the recommendation process in community question answering sites. We use Read2Vec, a readability assessment tool designed to measure the readability level of short documents, to inform a traditional content-based recommendation strategy. The results highlight the benefits of incorporating readability information in this process.

@en

The Internet is the biggest data-sharing platform, comprised of an immeasurable quantity of resources covering diverse topics appealing to users of all ages. Children shape tomorrow's society, so it is essential that this audience becomes agile with searching information. Although young users prefer well-known search engines, their lack of skill in formulating adequate queries and the fact that search tools were not designed explicitly with children in mind, can result in poor outcomes. The reasons for this include children's limited vocabulary, which makes it challenging to articulate information needs using short queries, or their tendency to create queries that are too long, which translates to few or irrelevant retrieved results. To enhance web search environments in response to children's behaviors and expectations, in this paper we discuss an initial effort to verify well-known issues, and identify yet to be explored ones, that affect children in formulating (natural language or keyword) queries. We also present a novel search intent module developed in response to these issues, which can seamlessly be integrated with existing search engines favored by children. The proposed module interprets a child's query and creates a shorter and more concise query to submit to a search engine, which can lead to a more successful search session. Initial experiments conducted using a sample of children queries validate the correctness of the proposed search intent module.

@en

We present our ongoing efforts on the development of a search environment tailored to 6-15 year-olds that can foster learning though retrieval of materials that not only satisfy the information needs of users but also match their reading abilities. You Understood.me is an enhanced environment based on a popular search engine specifically designed to help students deal with search for learning tasks, and allow teachers to track their progress. An initial assessment conducted on You Understood.me and well-known (children-oriented) search engines based on queries generated by K-9 students, showcases the need for this type of environment.

@en

We present an initial study examining the benefits of incorporating readability indicators in social network-related tasks. In order to do so, we introduce TweetRead, a readability assessment tool specifically designed for Twitter and use it to inform the hashtag prediction process, highlighting the importance of a readability signal in recommendation tasks.

@en