We have developed a novel hybrid representation for Music Information Retrieval. Our representation is built by incorporating audio content into the tag space in a tag-track matrix, and then learning hybrid concepts using latent semantic analysis. We apply this representation to the task of music recommendation, using similarity-based retrieval from a query music track. We also develop a new ap...

The main aim of a credit scoring model is the classification of the loan customers into two classes, reliable and unreliable customers, on the basis of their potential capability to keep up with their repayments. Nowadays, credit scoring models are increasingly in demand, due to the consumer credit growth. Such models are usually designed on the basis of the past loan applications and used to e...

Case Amplification can improve the accuracy of a collaborative filtering (CF) algorithm with no extra space overhead by amplifying the effect of close candidates in the prediction. However, in a cold start scenario, the traditional Case Amplification on an item-based prediction can reduce accuracy. Given a small known set, Case Amplification can give a mediocre candidate an unsuitable amplifica...

Sistemas de recomendação de filtragem colaborativa encontram diversos desafios para alcançar acuradas recomendações. Um deles é o denominado cold-start de item que é o fato do sistema não ser capaz de recomendar itens que nunca foram avaliados por outros usuários. Neste artigo apresentamos uma proposta para minimizar esse problema, o XPrefRec. Trata-se de um sistema de recomendação híbrido, seg...

Previous research showed that choice-based preference elicitation can be successfully used to reduce effort during user cold start, resulting in an improved user satisfaction with the recommender system. However, it has also been shown to result in highly popular recommendations. In the present study we investigate if trailers reduce this bias to popular recommendations by informing the user an...

In absence of individual user information, knowledge about larger user groups (e.g., country characteristics) can be exploited for deriving user preferences in order to provide recommendations to users. In this short paper, we study how to mitigate the cold-start problem on a country level for music retrieval. Specifically, we investigate a large-scale dataset on user listening behavior and sho...

The JHU HLTCOE participated in the Cold Start and the Entity Discovery and Linking tasks of the 2016 Text Analysis Conference Knowledge Base Population evaluation. For our fifth year of participation in Cold Start we continued our research with the KELVIN system. We submitted experimental variants that explore use of linking to Freebase across three languages and add relations beyond those requ...

This paper presents the system of the UGENT IBCN team for the TAC KBP 2014 slot filling and cold start (slot filling variant) tasks. This was the team’s first participation in both tasks. The slot filling system uses distant supervision to generate training data combined with a noise reduction step, and two different types of classifiers. We show that the noise reduction step significantly impr...

We propose a unified product embedded representation that is optimized for the task of retrieval-based product recommendation. We generate this representation using Content2Vec, a new deep architecture that merges product content information such as text and image, and we analyze its performance on hard recommendation setups such as cold-start and cross-category recommendations. In the case of ...

A goal for the creation and improvement of music recommendation is to retrieve users’ preferences and select the music adapting to the preferences. Although the existing researches achieved a certain degree of success and inspired future researches to get more progress, problem of the cold start recommendation and the limitation to the similar music have been pointed out. Hence we incorporate c...