Create accountLogin

Recent searches

No recent searches

Search options

Only available when logged in.
techhub.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
A hub primarily for passionate technologists, but everyone is welcome

Administered by:

Server stats:

4.9K
active users

techhub.social: AboutStatusProfiles directoryPrivacy policy

Mastodon: AboutGet the appKeyboard shortcutsView source codev4.3.8

#multimodalsearch

0 posts0 participants0 posts today
Public *
Mott

semion.io is an interesting multimodal search engine that retrieves academic articles from arxiv and displays side by side both an abstract of the article and figures from the article. It does allow you to do article search by text likely to occur in the illustrations/tables (or its caption), as each article is indexed in a directed graph with nodes representing each article component such individual figures, tables, equations, etc.

semion.io

semion.io dissects each manuscript into constituent components (paragraphs, tables, figures, chapters, equations...) and determines links, interconnections and citations. Internally, the semion search engine constructs a single directed graph of the combined information contained in all preprints on arXiv — the paper graph. Each node represents a figure, paragraph, equation or other constituent component and its associated description. Graph edges encode their interconnections, both within each original document and across paper boundaries. The task of retrieving results for a given search query thus maps onto a large-scale graph search problem. Under the hood, semion.io determines relevant partitions of the paper graph for each search request, then traverses the sub graph and computes a ranking of nodes that are deemed most relevant when placed in proper context.
#search#multimodalSearch
ExploreLive feeds
About