Examinando por Autor "Figueroa, Alejandro"
Mostrando 1 - 6 de 6
Resultados por página
Opciones de ordenación
Ítem Aprendizaje semi-supervisado para la detección de respuestas informativas en preguntas procedurales(Universidad Andres Bello, 2015) Palomera, Daniel.; Figueroa, Alejandro; Facultad de IngenieríaEs bueno mi poema? No soy bueno escribiendo poesía y tengo que leer este poema en la clase de mañana. Por favor ayudenme a arreglarlo. La procastinación se apodero de mi vida, Como un cuchillo, rebano mis calicaciones a la mitad. No voy a procrastinar nunca m as, repite mi mente, Pero todav a lo hago, Y la lista de todo lo que tengo que hacer crece. Twitter, Instagram y Snapchat necesitan ser desechados, Si quiero terminar este mes de trabajo atrasado. Tabla 1.1: Ejemplo de una pregunta realizada en Yahoo! Answers. Con el crecimiento de Internet, es cada vez m as dif cil encontrar la informaci on que estamos buscando, sobre todo cuando no sabemos la p agina exacta donde se encuentra. Una soluci on, por todos conocida, es el uso de motores de b usqueda (sitios en donde se busca informaci on en una gran cantidad de p aginas; como Google, Bing o Yahoo!) en donde escribimos, en pocas palabras, nuestra consulta. Pero esto es muy dif cil -o imposible- de realizar en algunos casos, en los que se puede recurrir a sitios especializados para realizar la pregunta y que, en base al conocimiento de otros usuarios, se les da respuesta. Por ejemplo, en Tabla 1.1, un usuario necesita que otras personas lo ayuden a mejorar el poema que ha escrito para una tarea escolar, pregunta que dif cilmente podr a ser resuelta en un motor de b usqueda. Estos tipos de sitios son llamados comunidades de pregunta-respuesta (a.k.a commu- nity question answering; cQA), en donde, a grandes rasgos, los usuarios pueden realizar preguntas de cualquier tipo y otros usuarios (algunos de ellos expertos en temas partic- ulares) las responden. En esta din amica, los usuarios que responden preguntas, suelen recibir puntos al ser seleccionada su respuesta como la mejor o al ser cali cada posi- tivamente por otros miembros de la comunidad. Otro aspecto que suelen tener estas comunidades son los votos del resto de los usuarios sobre las respuestas.Ítem Establecer una secuencia efectiva de aprendizaje a través de un modelo sistemático de búsqueda de construcción de datos(Universidad Andrés Bello, 2021) Contreras Mesa, Miguel Ángel; Llanos Sánchez, César; Vásquez Cariaga, Erwin; Figueroa, Alejandro; Ruete Zúñiga, David; Facultad de IngenieríaEn esta tesis, hemos establecido una secuencia efectiva de aprendizaje a través de un modelo sistemático de búsqueda de construcción de datos mediante la plataforma de Yahoo! Answers en el contexto de predicción de género y edad. Para obtener esta secuencia, se segmentaron los datos de entrenamiento según el género y año de nacimiento de cada usuario en lotes. Para seleccionar el mejor lote se implementó un algoritmo greedy considerando el mayor valor de la métrica macropromedio. Como resultado se obtuvo un curriculum o secuencia efectiva de aprendizaje, la cual fue verificada mediante el entrenamiento de un modelo de Bayes. Para comprobar el curriculum de aprendizaje se realizaron 2 experimentos, en el primer experimento se entrenó el modelo sin el curriculum, es decir, con todos los datos de los usuarios (329.025) ejemplos y en el segundo experimento se entrenó el modelo con el curriculum (11.176 ejemplos). El resultado final fue que ambos modelos tuvieron el mismo rendimiento de predicción, sin embargo, en el segundo experimento se necesitaron menos datos, por lo que su entrenamiento fue más rápido y eficiente, comprobando de esta manera la efectividad del curriculum.Ítem Improving Question Intent Identification by Exploiting Its Synergy With User Age(Institute of Electrical and Electronics Engineers Inc., 2023) Diaz, Octavio; Figueroa, AlejandroAt their heart, community Question-Answering (cQA) services are social networks that allow their members to prompt any kind of question expecting different answers produced by several community peers. Most of previous research on cQA has shown that questions can reflect two intents: learning information and starting a conversation. The purpose of this research is investigating the intrinsic relationship between models predicting question intent and user age. And if this relatedness can assist in overcoming one of the chief obstacles when constructing effective question intent recognizers: the scarcity of annotated data. The method adopted in this work involves addressing question intent recognition in a Multi-Task (MT) learning setting, where asker age identification is used as its auxiliary task. In other words, we exploit their task synergy by integrating both training signals with the aim of boosting the classification rate of question intent. Since MT learning is regarded as fruitful when a target task is improved wrt. single-task models, in our experiments, we compare four frontier frameworks with several state-of-the-art single-task neural network classifiers. In brief, our results show that a MT implementation of T5 yielded an increase of at least 10% over the best single-task models, when trained on full questions. Our experimental results also unveil that extra substantial improvements can be obtained by adjusting its parameters. All in all, we conclude that both variables are inherently related. Last but not least, we also make available a new question set labelled with the age of their askers and their intents with the hope of encouraging the research of MT learning into cQA tasks. © 2023 The Authors.Ítem Neural age screening on question answering communities(2023-08) Timilsina, Mohan; Figueroa, AlejandroFor online social networks, demographic analysis is absolutely essential for improving their services in many ways. It is instrumental in understanding their different audiences, members and competitors. As well as that, it is pivotal in designing effective personalization and contextualization strategies, especially for displaying and creating better content. There is, for this reason, a great bulk of research into how demographic variables are characterized and how they impact online platforms such as Facebook and Twitter. But surprisingly, only a handful of works delve into their characterization and effect on community Question-Answering (cQA) websites. In this particular context, the subject of age demographics remains largely unexplored. This paper takes the lead on interpreting automatic age recognition on CQAs (a.k.a. age screening) as a regression task. To this effect, it compares state-of-the-art graph-based neural network regression and embedding models on a massive activity-graph encompassing ca. 16 and 837 million nodes (members) and edges, respectively. For this study, a large-scale subset of ca. 657,000 community fellows was automatically associated with their age via aligning their profile texts with a limited number of linguistic patterns. In short, our results show that Node2vec significantly outperforms other embeddings regardless of the regression model used for casting predictions. When this embedding is combined with Artificial Neural Network Regressions, we obtained our best configuration scoring a Root Mean Square Error (RMSE) of 8.39. An interesting qualitative feature of this embedding space is that age-based centroid vectors tend to form a trail ordered by age. Lastly, our outcomes also signal that activity graph based models can rival its counterparts based on image and textual inputs, paving the way for constructing effective multi-modal approaches. © 2023 The Author(s)Ítem Search clicks analysis for discovering temporally anchored questions in community Question Answering(Elsevier Ltd, 2016-05) Figueroa, Alejandro; Gómez-Pantoja, Carlos; Herrera, IgnacioNowadays, community Question-Answering (cQA) sites are massive repositories for user-generated content, where members prompt questions expecting satisfactory answers from other members. However, in this dynamic, there is an intrinsic delay between the moment questions are posted to the arrival of acceptable responses. Therefore, cQA platforms have the pressing need for promoting unresolved questions to potential answerers and for taking advantage of resolved questions contained in their archives, whenever possible. This paper studies cQA services from the viewpoint of the time frame where their questions attract the interest of their community members. By drawing a parallel with temporal patterns of user interests in web search activity, we are able to define three main types of temporally anchored questions: trend or bursty, periodic and permanent. Then, by analyzing user click distributions to Yahoo! Answers pages across Yahoo! Search logs, we automatically acquired a set of 35,000 cQA questions labeled with one of these three temporal anchors. Accordingly, we show the practicality of this approach by means of human assessments; and by using this automatically acquired corpus for studying several classification models. Essentially, the proposed method was found to correlate well with these human judgements, and proven to be effective in building systems that automatically identify the temporal anchor of unseen cQA questions. In substance, our outcomes indicate that some contexts are strongly related to a particular temporal anchor. We believe that these anchors will contribute to the discrimination of resolved questions that are capable of being revitalized, as well as to foster the opportune participation in questions that generate enthusiasm only for a short time. © 2015 Elsevier Ltd. All rights reserved.Ítem Three-dimensional assessment of Upper Airway in Class III patients with different facial patterns(Medicina Oral S.L., 2023) De Nordenflycht, Diego; Corona, Tomás; Figueroa, AlejandroBackground: To evaluate three-dimensionally the upper airway (UA) of class III adults with different facial patterns. Material and Methods: a cross-sectional study was conducted, in which cone-beam computed tomography (CBCT) images from a private clinic in Viña del Mar, Chile were evaluated. The sample consisted of CBCT images of 59 skeletal class III subjects (33 females and 26 males, mean age 24.7 years) in which the vertical facial pattern was determined using the Vert index, and the minimum cross-sectional area and total volume of the UA were measured. The minimum cross-sectional area variable was analyzed by ANOVA and the total volume was analyzed by Kruskal- Wallis test. Statistical analyses were performed with JASP 0.13.1 software at p=0.05. Results: the sample included images of 21 brachyfacial, 14 mesofacial and 24 dolichofacial subjects. The mean minimum cross-sectional area of the sample was 591.78 mm2 +/- 149.38 mm2 (minimum=352.00 mm2; maximum= 971.00 mm2), being greater in brachyfacial than in dolichofacial and mesofacial subjects, however, these differences were not significant (p=0.147). The mean total volume of the sample was 13.40 +/- 4.69 cm3 (minimum= 7.16 cm3; maximum=25.66 cm3), being greater in brachyfacial than in dolichofacial and mesofacial subjects, however, these differences were not significant (p=0.353). Conclusions: Considering the limitations of the present study, the vertical facial pattern does not appear to significantly influence upper airway measurements in skeletal class III adults. © Medicina Oral S. L.