SocioFillmore

Computational Frame Semantics for Events in Society

When using language, we talk about events differently depending on where we find ourselves in space, time, society, and ideology.

Did I lend you a book or did you borrow a book from me?
Did planes hit the WTC or do we remember the 9/11 attacks?
Did a driver injure a child or did a cyclist collide with a car?
Did a man murder a woman or did a tragic incident occur?

This NLP tool based on frame semantics and construction grammar can help visualize perspective shifts caused by backgrounding of event participants and their agentivity.

How would you like to use the app?
Interactive Mode Explore Existing Datasets

Interactive Mode

First, define the real-world event type that you would like to analyze.

Show me an example

Event type

Next, define a set of frames — pieces of conceptual information the define how language users imagine and describe your event. If you're not sure which frames to use, we can help you find relevant frames either using a keyword search or by automatically processing a text that you provide.

Show me an example

Semantic Frames

Select up to three frames that define your event. Where applicable, we will automatically detect further frames that provide alternative perspectives on your event.

@{idx + 1}
Main Perspective
@{frame.name}

@{frame.typicalExemplar.text}
@{altFrame.type} Perspective
@{altFrame.name}

@{altFrame.typicalExemplar.text}
Add frames manually
Help me choose frames
Infer from keywords

To help you choose frames, please enter some keywords describing your event:

Keywords

No keywords selected yet.

One or more of the keywords you selected are invalid.

@{kwd.text}

(Click a keyword to remove it)

Suggested frames

Based on your keywords, we suggest the frames listed below.

Frames marked with the eye symbol ("") are more likely to yield interesting results because they allow for automatically detecting alternative perspectives on the event.

@{frm.text}

Examples
...@{ex.text}...

Enter the text you want to analyze below.

Show me an example

Text Analysis

Powered by LOME (Xia et al. EACL 2021)

Enter any text in Italian, Dutch, or English.

sentence @{sentIdx + 1}
@{part.text} @{part.text} @{part.text} @{part.text}