Dig into Multi-modal Cues for Video Retrieval with Hierarchical Alignment

Multi-modal cues presented in videos are usually beneficial for the challenging video-text retrieval task on internet-scale datasets. Recent video retrieval methods take advantage of multi-modal cues by aggregating them to holistic high-level semantics for matching with text representations in a global view. In contrast to this global alignment, the local alignment of detailed semantics encoded within both multi-modal cues and distinct phrases is still not well conducted. Thus, in this paper, we leverage the hierarchical video-text alignment to fully explore the detailed diverse characteristics in multi-modal cues for fine-grained alignment with local semantics from phrases, as well as to capture a high-level semantic correspondence. Specifically, multi-step attention is learned for progressively comprehensive local alignment and a holistic transformer is utilized to summarize multi-modal cues for global alignment. With hierarchical alignment, our model outperforms state-of-the-art methods on three public video retrieval datasets.

THE VEDIC «FEET» OF THE GREEK SPHINX

Object of the article: the riddle of the Sphinx and the Vedic riddles. Subject of the article: structural and semantic comparison between the riddle of the Sphinx and the riddles in “Rigveda” and “Yajurveda”. Purpose of research: to show that the scheme of the riddle of the Sphinx about “man”, based on the description of an allegorical creature which is “two-footed, three-footed and fourfooted”, has a structural correspondence in the Vedic ritual-offering formulas, which describe Vedic poetic metres, divided by number of padas (“feet”) in a verse. Research methods: methods of the structural and semantic analysis applied. Results: the author argues that the poetic scheme of the Vedic ritual-offering formulas reflects an Indo-European proto-text, on which the riddle of the Sphinx is based. Field of application: literary studies. Conclusion:The author argues that the description of the Vedic metres in ritual-offering formulas through the allegorical ”two-footed (dvipada), four-footed (catushpada) and three-footed (tripada)”creatures has a structural and semantic correspondence in the riddle of the Sphinx, based on the description of an allegorical creature which is “two-footed, three-footed and four-footed”. The essay suggests that this correspondence between the Vedic formulas and the riddle of the Sphinx allows an alternative interpretation of the Greek riddle on meta-literal level as an allegorical description of poetic forms. The author is indebted to Svetlana Gracheva and Dan Whitman for stimulating discussions and helpful advice.

What Is the Sense in Logic and Philosophy of Language

In the paper, various notions of the logical semiotic sense of linguistic expressions – namely, syntactic and semantic, intensional and extensional – are considered and formalised on the basis of a formal-logical conception of any language L characterised categorially in the spirit of certain Husserl's ideas of pure grammar, Leśniewski-Ajdukiewicz's theory of syntactic/semantic categories and, in accordance with Frege's ontological canons, Bocheński's and some of Suszko's ideas of language adequacy of expressions of L. The adequacy ensures their unambiguous syntactic and semantic senses and mutual, syntactic and semantic correspondence guaranteed by the acceptance of a postulate of categorial compatibility of syntactic and semantic (extensional and intensional) categories of expressions of L. This postulate defines the unification of these three logical senses. There are three principles of compositionality which follow from this postulate: one syntactic and two semantic ones already known to Frege. They are treated as conditions of homomorphism of partial algebra of L into algebraic models of L: syntactic, intensional and extensional. In the paper, they are applied to some expressions with quantifiers. Language adequacy connected with the logical senses described in the logical conception of language L is, obviously, an idealisation. The syntactic and semantic unambiguity of its expressions is not, of course, a feature of natural languages, but every syntactically and semantically ambiguous expression of such languages may be treated as a schema representing all of its interpretations that are unambiguous expressions.