Support for Explicit Term Selection and Expansion in SPLADE

[celsofranssa]

Does SPLADE support selecting representative terms from a text and expanding them into related terms as shown in the following figure?​

For example, given a sentence like "Do any large monkeys come from the jungles of Indonesia," can SPLADE:

1. Identify representative terms such as:

"large monkeys," "come from the," "jungles," and "Indonesia."​

2. Expand each selected term into related terms, resulting in:

"large monkeys" → ["orangutans", "great apes", "gorillas"]​
"come from the" → ["native"]​
"jungles" → ["lands", "rainforest"]​
"Indonesia" → ["Borneo", "Malaysia", "Java"]​

If SPLADE does not directly support this functionality, are there recommended approaches or integrations to achieve similar term selection and expansion?