6.7 Future Work

While the presented corpus is currently usable (as demonstrated in our real use-case application described in Sevilla, Díaz, y Lahoz-Bengoechea 2023), it represents only an initial step in its development.

There remains a significant portion of the corpus to be annotated, and this further annotation might uncover issues that require some revision of the annotation schema.

Additionally to expanding the coverage of the annotation, the data collected can be augmented along two axes. Collecting data from more informants will make the corpus more robust and representative of actual SignWriting. Collecting data from continuous, real-world usage, instead of isolated examples, may also further this goal.

On a different note, these last years have witnessed significant advancements in deep learning approaches. New, state-of-the-art algorithms may be able to tell us more about our collected data, expedite the annotation process, or maybe even open up avenues for exploitation which we had not thought about before.

Finally, it is our hope that the public availability of the corpus will invite other researchers to contribute to this future work or add their own insights. The involvement of a broader research community will surely lead to a richer understanding of SignWriting or even sign languages themselves.

{
  "meta": {
    "glosa": "Mentira"
  },
  "fold": 8,
  "graphemes": [
    {
      "tags": {
        "CLASS": "HEAD", "SHAPE": "cheekr"
      },
      "box": [ 0.3561, 0.3867, 0.3992, 0.4156 ]
    },
    {
      "tags": {
        "CLASS": "DIAC", "SHAPE": "rub"
      },
      "box": [ 0.5134, 0.1207, 0.0944, 0.1507 ]
    },
    {
      "tags": {
        "CLASS": "ARRO", "SHAPE": "b",
        "ROT": "N"
      },
      "box": [ 0.6032, 0.1532, 0.1159, 0.1145 ]
    },
    {
      "tags": {
        "CLASS": "STEM", "SHAPE": "s",
        "ROT": "N"
      },
      "box": [ 0.6038, 0.3787, 0.0619, 0.25 ]
    },
    {
      "tags": {
        "CLASS": "HAND", "SHAPE": "I", "VAR": "b",
        "REF": "y", "ROT": "N"
      },
      "box": [ 0.7224, 0.5644, 0.1913, 0.3229 ]
    },
    {
      "tags": {
        "CLASS": "ARRO", "SHAPE": "b",
        "ROT": "N"
      },
      "box": [ 0.6025, 0.2363, 0.1078, 0.0763 ]
    }
  ]
}