Ionosphere.tv
feat: define tv.ionosphere.nlpAnnotations lexicon for lens source schema
Formalizes the NLP pipeline JSON output shape so panproto can use it as
a typed lens source schema. Not published to PDS — exists purely to
enable cross-schema lens derivation from NLP annotations.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
+148
+148
formats/tv.ionosphere/nlpAnnotations.lexicon.json
+148
formats/tv.ionosphere/nlpAnnotations.lexicon.json
···
1
+
{
2
+
"lexicon": 1,
3
+
"$type": "com.atproto.lexicon.schema",
4
+
"id": "tv.ionosphere.nlpAnnotations",
5
+
"revision": 1,
6
+
"description": "NLP pipeline output for a conference talk transcript. Formalizes the JSON shape produced by the ionosphere NLP pipeline so panproto can use it as a lens source schema. Not published to PDS.",
7
+
"defs": {
8
+
"main": {
9
+
"type": "record",
10
+
"key": "any",
11
+
"record": {
12
+
"type": "object",
13
+
"required": ["talkRkey", "sentences", "paragraphs", "entities", "topicBreaks", "metadata"],
14
+
"properties": {
15
+
"talkRkey": {
16
+
"type": "string",
17
+
"description": "The rkey of the tv.ionosphere.talk record this annotation set covers."
18
+
},
19
+
"sentences": {
20
+
"type": "array",
21
+
"items": {
22
+
"type": "ref",
23
+
"ref": "tv.ionosphere.nlpAnnotations#nlpSentence"
24
+
},
25
+
"description": "Sentence-boundary spans detected in the transcript text."
26
+
},
27
+
"paragraphs": {
28
+
"type": "array",
29
+
"items": {
30
+
"type": "ref",
31
+
"ref": "tv.ionosphere.nlpAnnotations#nlpParagraph"
32
+
},
33
+
"description": "Paragraph-boundary spans detected in the transcript text."
34
+
},
35
+
"entities": {
36
+
"type": "array",
37
+
"items": {
38
+
"type": "ref",
39
+
"ref": "tv.ionosphere.nlpAnnotations#nlpEntity"
40
+
},
41
+
"description": "Named entities detected in the transcript text."
42
+
},
43
+
"topicBreaks": {
44
+
"type": "array",
45
+
"items": {
46
+
"type": "ref",
47
+
"ref": "tv.ionosphere.nlpAnnotations#nlpTopicBreak"
48
+
},
49
+
"description": "Byte positions where a topic shift was detected."
50
+
},
51
+
"metadata": {
52
+
"type": "ref",
53
+
"ref": "tv.ionosphere.nlpAnnotations#nlpMetadata",
54
+
"description": "Provenance and configuration metadata for this annotation run."
55
+
}
56
+
}
57
+
}
58
+
},
59
+
"nlpSentence": {
60
+
"type": "object",
61
+
"description": "A sentence span identified by the NLP pipeline.",
62
+
"required": ["byteStart", "byteEnd"],
63
+
"properties": {
64
+
"byteStart": {
65
+
"type": "integer",
66
+
"description": "Inclusive start byte offset into the transcript text."
67
+
},
68
+
"byteEnd": {
69
+
"type": "integer",
70
+
"description": "Exclusive end byte offset into the transcript text."
71
+
}
72
+
}
73
+
},
74
+
"nlpParagraph": {
75
+
"type": "object",
76
+
"description": "A paragraph span identified by the NLP pipeline.",
77
+
"required": ["byteStart", "byteEnd"],
78
+
"properties": {
79
+
"byteStart": {
80
+
"type": "integer",
81
+
"description": "Inclusive start byte offset into the transcript text."
82
+
},
83
+
"byteEnd": {
84
+
"type": "integer",
85
+
"description": "Exclusive end byte offset into the transcript text."
86
+
}
87
+
}
88
+
},
89
+
"nlpEntity": {
90
+
"type": "object",
91
+
"description": "A named entity span identified by the NLP pipeline.",
92
+
"required": ["byteStart", "byteEnd", "label", "nerType"],
93
+
"properties": {
94
+
"byteStart": {
95
+
"type": "integer",
96
+
"description": "Inclusive start byte offset into the transcript text."
97
+
},
98
+
"byteEnd": {
99
+
"type": "integer",
100
+
"description": "Exclusive end byte offset into the transcript text."
101
+
},
102
+
"label": {
103
+
"type": "string",
104
+
"description": "Surface text of the named entity (e.g. 'Matt')."
105
+
},
106
+
"nerType": {
107
+
"type": "string",
108
+
"description": "NER category label (e.g. PERSON, ORG, GPE)."
109
+
},
110
+
"conceptUri": {
111
+
"type": "string",
112
+
"format": "at-uri",
113
+
"description": "AT URI of a canonical concept record for this entity, if resolved."
114
+
}
115
+
}
116
+
},
117
+
"nlpTopicBreak": {
118
+
"type": "object",
119
+
"description": "A position in the transcript text where the NLP pipeline detected a topic shift.",
120
+
"required": ["byteStart"],
121
+
"properties": {
122
+
"byteStart": {
123
+
"type": "integer",
124
+
"description": "Byte offset at which the topic shift begins."
125
+
}
126
+
}
127
+
},
128
+
"nlpMetadata": {
129
+
"type": "object",
130
+
"description": "Provenance and configuration metadata for an NLP annotation run.",
131
+
"required": ["tool"],
132
+
"properties": {
133
+
"tool": {
134
+
"type": "string",
135
+
"description": "Name and version of the NLP tool used (e.g. 'spacy/en_core_web_sm')."
136
+
},
137
+
"pauseThresholdMs": {
138
+
"type": "integer",
139
+
"description": "Silence duration in milliseconds used to delimit paragraphs."
140
+
},
141
+
"proximityWords": {
142
+
"type": "integer",
143
+
"description": "Word-window radius used when linking entity mentions to concepts."
144
+
}
145
+
}
146
+
}
147
+
}
148
+
}