Table 1.

Descriptions and examples of Snowball objects.

Object name	Description Example
Sentence	Sentence split from body text. “The bulk TiO2 has a direct band gap of 3.2 eV at tau point.”
Relationship tuple	An entity list made of chemical name, property specifier, property value, and property unit. [TiO2, band gap, 3.2, eV]
Prefix	Word tokens before the first entity. [“The”, “bulk”]
Middle	Word tokens between each relationship entity. [“has”, “a”, “direct”, “of”]
Suffix	Word tokens after the last entity. [“at”, “tau”, “point”]
Weight	Normalized importance factor for prefix, middles, and suffix. (0.1, 0.8, 0.1)
Phrase object	A normalized vector of entities. [“The bulk”, chemical name, “has”, “a”, “direct”, specifier, “of”, value, unit, “at tau point”]