Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add delimited term frequency token filter documentation #5043

Merged
merged 11 commits into from
Sep 22, 2023
275 changes: 275 additions & 0 deletions _analyzers/token-filters/delimited-term-frequency.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,275 @@
---
layout: default
title: Delimited term frequency
parent: Token filters
nav_order: 100
---

# Delimited term frequency token filter

The `delimited_term_freq` token filter separates a token stream into tokens with corresponding term frequencies, based on a provided delimiter. A token consists of all characters before the delimiter, and a term frequency is the integer after the delimiter. For example, if the delimiter is `|`, then for the string `foo|5`, `foo` is the token and `5` is its term frequency. If there is no delimiter, the token filter does not modify the term frequency.

You can either use a preconfigured `delimited_term_freq` token filter or create a custom one.

## Preconfigured `delimited_term_freq` token filter

The preconfigured `delimited_term_freq` token filter uses the `|` default delimiter. To analyze text with the preconfigured token filter, send the following request to the `_analyze` endpoint:

```json
POST /_analyze
{
"text": "foo|100",
"tokenizer": "keyword",
"filter": ["delimited_term_freq"],
"attributes": ["termFrequency"],
"explain": true
}
```
{% include copy-curl.html %}

The `attributes` array specifies that you want to filter the output of the `explain` parameter to return only `termFrequency`. The response contains both the original token and the parsed output of the token filter that includes the term frequency:

```json
{
"detail": {
"custom_analyzer": true,
"charfilters": [],
"tokenizer": {
"name": "keyword",
"tokens": [
{
"token": "foo|100",
"start_offset": 0,
"end_offset": 7,
"type": "word",
"position": 0,
"termFrequency": 1
}
]
},
"tokenfilters": [
{
"name": "delimited_term_freq",
"tokens": [
{
"token": "foo",
"start_offset": 0,
"end_offset": 7,
"type": "word",
"position": 0,
"termFrequency": 100
}
]
}
]
}
}
```

## Custom `delimited_term_freq` token filter

To configure a custom `delimited_term_freq` token filter, first specify the delimiter in the mapping request, in this example, `^`:

```json
PUT /testindex
{
"settings": {
"analysis": {
"filter": {
"my_delimited_term_freq": {
"type": "delimited_term_freq",
"delimiter": "^"
}
}
}
}
}
```
{% include copy-curl.html %}

Then analyze text with the custom token filter you created:

```json
POST /testindex/_analyze
{
"text": "foo^3",
"tokenizer": "keyword",
"filter": ["my_delimited_term_freq"],
"attributes": ["termFrequency"],
"explain": true
}
```
{% include copy-curl.html %}

The response contains both the original token and the parsed version with the term frequency:

```json
{
"detail": {
"custom_analyzer": true,
"charfilters": [],
"tokenizer": {
"name": "keyword",
"tokens": [
{
"token": "foo|100",
"start_offset": 0,
"end_offset": 7,
"type": "word",
"position": 0,
"termFrequency": 1
}
]
},
"tokenfilters": [
{
"name": "delimited_term_freq",
"tokens": [
{
"token": "foo",
"start_offset": 0,
"end_offset": 7,
"type": "word",
"position": 0,
"termFrequency": 100
}
]
}
]
}
}
```

## Combining `delimited_token_filter` with scripts

You can write Painless scripts to calculate custom scores for the documents in the results.

First, create an index and provide the following mappings and settings:

```json
PUT /test
{
"settings": {
"number_of_shards": 1,
"analysis": {
"tokenizer": {
"keyword_tokenizer": {
"type": "keyword"
}
},
"filter": {
"my_delimited_term_freq": {
"type": "delimited_term_freq",
"delimiter": "^"
}
},
"analyzer": {
"custom_delimited_analyzer": {
"tokenizer": "keyword_tokenizer",
"filter": ["my_delimited_term_freq"]
}
}
}
},
"mappings": {
"properties": {
"f1": {
"type": "keyword"
},
"f2": {
"type": "text",
"analyzer": "custom_delimited_analyzer",
"index_options": "freqs"
}
}
}
}
```
{% include copy-curl.html %}

The `test` index uses a keyword tokenizer, a delimited term frequency token filter (where the delimiter is `^`), and a custom analyzer that includes a keyword tokenizer and a delimited term frequency token filter. The mappings specify that the field `f1` is a keyword field and the field `f2` is a text field. The field `f2` uses the custom analyzer defined in the settings for text analysis. Additionally, specifying `index_options` signals to OpenSearch to add the term frequencies to the inverted index. You'll use the term frequencies to give documents with repeated terms a higher score.

Next, index two documents using bulk upload:

```json
POST /_bulk?refresh=true
{"index": {"_index": "test", "_id": "doc1"}}
{"f1": "v0|100", "f2": "v1^30"}
{"index": {"_index": "test", "_id": "doc2"}}
{"f2": "v2|100"}
```
{% include copy-curl.html %}

The following query searches for all documents in the index and calculates document scores as the term frequency of the term `v1` in the field `f2`:

```json
GET /test/_search
{
"query": {
"function_score": {
"query": {
"match_all": {}
},
"script_score": {
"script": {
"source": "termFreq(params.field, params.term)",
"params": {
"field": "f2",
"term": "v1"
}
}
}
}
}
}
```
{% include copy-curl.html %}

In the response, document 1 has a score of 30 because the term frequency of the term `v1` in the field `f2` is 30. Document 2 has a score of 0 because the term `v1` does not appear in `f2`:
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should the first instance of "document" be capitalized?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think so because it's not a proper name of the document?


```json
{
"took": 4,
"timed_out": false,
"_shards": {
"total": 1,
"successful": 1,
"skipped": 0,
"failed": 0
},
"hits": {
"total": {
"value": 2,
"relation": "eq"
},
"max_score": 30,
"hits": [
{
"_index": "test",
"_id": "doc1",
"_score": 30,
"_source": {
"f1": "v0|100",
"f2": "v1^30"
}
},
{
"_index": "test",
"_id": "doc2",
"_score": 0,
"_source": {
"f2": "v2|100"
}
}
]
}
}
```

## Parameters

The following table lists all parameters that the `delimited_term_freq` supports.

Parameter | Required/Optional | Description
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I like how you styled this heading. I'll follow same format.

:--- | :--- | :---
`delimiter` | Optional | The delimiter used to separate tokens from term frequencies. Must be a single non-null character. Default is `|`.
Loading
Loading