In Combinatory Categorial Grammar (Steedman 1996, 2000), questions and focused sentences have been considered to be categories of S. I claim that these types of sentences should be functions from a sentence to another sentence in view of their semantics. I analyze Japanese sentence-final particles as verum focus operators or question markers whose categories are S\(S\S).

In a costumers' network obtained, for example, from social networks, modeling influences among buyers is useful to estimate a buyer's decision to buy an item. This kind of influences is sometimes modeled with submodular functions and the problem of finding the marketing strategies to offer items to buyers that maximize the revenues of other buyers can be formulated as submodular maximization. In this paper, we address the problem of enumerating optimal and near-optimal solutions (strategies) to this problem and investigate numerically with real social network data.

A physical system can be represented by autoregressive and moving average (ARMA) model in discrete time domain, which can determine system's parameters. We extend the ARMA model to an ARMA-LiNGM model which is a canonical form to represent the physical system. We further analyze the relationships between the ARMA-LiNGM model and a CTARMA model which is another canonical form in continuous time domain. As the consequence, we provide the relations between the coefficients of the two models.