sentence.proto

syntax = "proto2";
option java_package = "ml.generall";

// Position of meta info
message Position {
    // from position
    required int32 fromPos = 1;
    // to position
    required int32 toPos = 2;
    // Next available id: 3
}

// parameters of mention
message Params {
    // Summ weight of tokens
    optional double sum_weight = 1; // Required
    // average weight of tokens
    optional double avg_wight = 2;  // Required
    // Max weight of tokens
    optional double max_wight = 3;  // Required
    // count of tokens
    optional int32 word_count = 4;  // Required
    // Next available id: 5
}

// Context of mention
message Context {
    // size of context
    required int32 size = 1;  // Required
    // left context string
    optional string left = 2;
    // right context string
    optional string right = 3;
    // Next available id: 1
}

// Concept link
message Concept {
    // Link to concept
    required string link = 1;  // Required
    // Count of hits
    optional int32 hits = 2;

    // Average score
    optional double avgScore = 3;

    // Max score
    optional double maxScore = 4;

    // Minimal score
    optional double minScore = 5;

    // Average normalized score
    optional double avgNorm = 6;

    // Average soft-max normalized score
    optional double avgSoftMax = 7;

    // Next available id: 8
}

//Mention description 
message Mention {
    // id of mention
    required int32 id = 1;  // Required int64 id = 1;
    // Resolver name: elastic or wikimention
    optional string resolver = 2;  // Required

    // Text of mention
    optional string text = 3;  // Required

    // Position of mention
    optional Position position = 4;  // Required

    // Params of mention
    optional Params params = 5;  // Required

    // Context of mention
    optional Context context = 6;  // Required

    // Concepts
    repeated Concept concepts = 7;

    // Next available id: 8
}

// Token description
message Token {
    // Token string
    optional string token = 1;  // Required

    // Lemma of token
    optional string lemma = 2;  // Required

    // Part of speech tag
    optional string pos_tag = 3;  // Required

    // parser tag
    optional string parserTag = 4;  // Required

    // Group id of token
    optional int32 group = 5;  // Required

    // Mentions id of this token
    repeated int32 mentions = 6;

    // Next available id: 1
}

// Sentence object. 
message Sentence {
    // Complete sentence text
    optional string sent = 1;  // Required

    // Mentions
    repeated Mention mentions = 2;

    // Name of sentence parser: CoreNLP, OpenNLP e.t.c
    optional string parser_name = 3;  // Required


    // Result of string parsing
    repeated Token parse_result = 4;

    // Sentence before observed
    optional string prevSentence = 5;

    // Sentence after observed
    optional string nextSentence = 6;


    // Next available id: 7
}