Josep Lluís de la Rosa i Esteva

Asknext: An agent protocol for social search

This paper describes the Asknext protocol, which automates knowledge exchanges by connecting agents using social networks. The protocol combines social answering systems, in which people answer queries or questions, with social feedback systems, in which feedback is used to sort search results. In Asknext, each agent represents a user, her contacts and her knowledge. When an agent receives a question, it will try to answer it; if it has no answer available, the agent forwards a message with that question to its contacts or shows it to its user. Asknexts main contributions are that it introduces stop messages and different propagation speeds for different messages, which allow a reduction in the spread of messages after an information need has been satisfied. A difference in the propagation speed of messages results in the search stopping when a relevant answer has been found. This paper describes a formalisation of the protocol, its mathematical model and experiments showing that the number of messages generated is lower than those generated by social search protocols such as Sixearch. Concretely, Asknext sends 20 times fewer messages than Sixearch; therefore, Asknext significantly improves the scalability of social search protocols, with as little as 0.05% loss of answer suitability, with the result that in many cases, there is no loss of suitability.

Survey of social search from the perspectives of the village paradigm and online social networks

Two paradigms currently exist for information search. The first is the library paradigm, which has been largely automated and is the prevailing paradigm in today’s web search. The second is the village paradigm, and although it is older than the library paradigm, its automation has not been considered, yet certain elements of its key aspects have been automated, as in the cases of the Q&A communities or novel services such as Quora. The increasing popularity and availability of online social networks and question-answering communities have encouraged revisiting of the automation of the village paradigm owing to new helpful developments, primarily that people are more connected with their acquaintances on the internet and their contact lists are available. In this survey, we study how the village paradigm is today partially automated: we consider the selection of candidates for answering questions, answering questions automatically and helping candidates to decide what questions to answer. Other aspects are also considered, for example, the automation of a reward system. We conclude that a next step towards the automation of the village paradigm involves intelligent agents that can leverage a P2P (peer-to-peer) social network, which will create new and interesting issues deeply entwined with social networks in the form of information processing by agents in parallel and side by side with people.Two paradigms currently exist for information search. The first is the library paradigm, which has been largely automated and is the prevailing paradigm in todays web search. The second is the village paradigm, and although it is older than the library paradigm, its automation has not been considered, yet certain elements of its key aspects have been automated, as in the cases of the Q&A communities or novel services such as Quora. The increasing popularity and availability of online social networks and question-answering communities have encouraged revisiting of the automation of the village paradigm owing to new helpful developments, primarily that people are more connected with their acquaintances on the internet and their contact lists are available. In this survey, we study how the village paradigm is today partially automated: we consider the selection of candidates for answering questions, answering questions automatically and helping candidates to decide what questions to answer. Other aspects are also considered, for example, the automation of a reward system. We conclude that a next step towards the automation of the village paradigm involves intelligent agents that can leverage a P2P (peer-to-peer) social network, which will create new and interesting issues deeply entwined with social networks in the form of information processing by agents in parallel and side by side with people.

Question Waves: A multicast query routing algorithm for social search

This paper describes Question Waves, an algorithm that can be applied to social search protocols, such as Asknext or Sixearch. In this model, the queries are propagated through the social network, with faster propagation through more trustable acquaintances. Question Waves uses local information to make decisions and obtain an answer ranking. With Question Waves, the answers that arrive first are the most likely to be relevant, and we computed the correlation of answer relevance with the order of arrival to demonstrate this result. We obtained correlations equivalent to the heuristics that use global knowledge, such as profile similarity among users or the expertise value of an agent. Because Question Waves is compatible with the social search protocol Asknext, it is possible to stop a search when enough relevant answers have been found; additionally, stopping the search early only introduces a minimal risk of not obtaining the best possible answer. Furthermore, Question Waves does not require a re-ranking algorithm because the results arrive sorted.

Propagation of question waves by means of trust in a social network

This paper proposes a model of behavior for agents who answer questions; the model works similarly to the way in which people interact in social networks and the agents behave differently depending on who asks a question; this behavior modulates the effort utilized in finding better answers for a given question. Our model also avoids consulting all acquaintances fact that can overload or overburden the contacts. However, since reducing the number of recipients might result in a poorer answer, we propose a behavior of consulting a small set of contacts and adding more recipients only if no relevant answer is found. The most promising result is that the first answer-in is probably the most relevant. The ordering the answers simply as they arrive gives the best ranking of answers. The new ranking is well-suited for real time question answering and avoids costly methods associated with re-ranking results.

Applying Short-term Memory to Social Search Agents☆

Abstract This paper presents about our research in social search. Generally, the research in social search falls into two principal challenges. The first challenge is how to find more relevant answers to the question. The second one is how to increase speed in finding relevant answers. Recently, we had provided two algorithms called Asknext and Question Waves to find more relevant answers compared to the baseline algorithm BFS. But, the search speed of the two proposed algorithms still the subject to improve. In this paper, we introduce the agents’ ability of learning the answers from the interactions with other agents so that they can quickly answer the question of other agents. We model this learning process by implementing the concept of data caching as the short-term memory of each social search agent. The result improvement of the speediness and the reduction of the number of messages used to communicate between agents, after apply agents short-term memory concept, demonstrates the usefulness of the proposed approach.

Naming Game Dynamics on Pairs of Connected Networks with Competing Opinions

We study the Naming Game (NG) dynamics when two disjoint networks with nodes in consensus on competing opinions are connected with new links. We consider two sets of networks; one contains several networks with real-life communities, the other networks generated with the Watts-Strogatz and Barabasi-Albert models. For each set, we run NG on all the possible pairs of networks and observe whether a consensus is reached to determine network features that correlate highly with such outcome. The main conclusion is that the quality of network community structure informs network’s ability to resist or exert influence from/on others. Moreover, the outcomes depend on whether Speaker-First of Listener-First NG is run and on whether a speaker or listener is biased towards high or low degree nodes. The results reveal strategies that may be used to enable and accelerate convergence to consensus in social networks.

Discovering Good Sources for Recommender Systems

Discovering user knowledge is a key issue in recommender systems and many algorithms and techniques have been used in the attempt. One of the most critical problems in recommender systems is the lack of information, referred to as Cold Start and Sparsity problems. Research works have shown how to take advantage of additional databases with information about users [1], but they do not solve the new problem that arises: which relevant database to use? This paper contributes to that solution with a novel method for selecting information sources in the belief that they will be relevant and will result in better recommendations. We describe a new approach to explore and discover relevant information sources in order to obtain reliable knowledge about users. The relation between the improvement of the recommendation results and the sources selected based on these characteristics is shown by experiments selecting source based on their relevance and trustworthiness.