[jira] [Commented] (SOLR-7193) Concatenate words from token stream
[ https://issues.apache.org/jira/browse/SOLR-7193?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16508806#comment-16508806 ] Alexandre Rafalovitch commented on SOLR-7193: - This seems to be satisfied by LUCENE-8332 and SOLR-12376, both coming in 7.4. > Concatenate words from token stream > --- > > Key: SOLR-7193 > URL: https://issues.apache.org/jira/browse/SOLR-7193 > Project: Solr > Issue Type: New Feature > Components: Schema and Analysis >Reporter: Abhishek Bafna >Priority: Major > Attachments: concatenate_words.patch > > > The user entered data often don't have proper spacing between words and words > spelling and format also varies from data like business names, address etc. > After tokenizing data, we might perform pattern replacement, stop word > filtering etc. Later we want to concatenate all the tokens and generate > n-grams token for indexing business name and perform the fuzzy match. -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-7193) Concatenate words from token stream
[ https://issues.apache.org/jira/browse/SOLR-7193?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14359939#comment-14359939 ] abhishek bafna commented on SOLR-7193: -- [~jmtd890917] Did you get the point I tried to convey. Can you please provide your further comment for the patch. Concatenate words from token stream --- Key: SOLR-7193 URL: https://issues.apache.org/jira/browse/SOLR-7193 Project: Solr Issue Type: New Feature Components: Schema and Analysis Reporter: abhishek bafna Attachments: concatenate_words.patch The user entered data often don't have proper spacing between words and words spelling and format also varies from data like business names, address etc. After tokenizing data, we might perform pattern replacement, stop word filtering etc. Later we want to concatenate all the tokens and generate n-grams token for indexing business name and perform the fuzzy match. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-7193) Concatenate words from token stream
[ https://issues.apache.org/jira/browse/SOLR-7193?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14358173#comment-14358173 ] abhishek bafna commented on SOLR-7193: -- The ConcatenateWordsFilter takes all the input token (words) and generate a single token. The CPU time and memory depends on the number and size of the tokens coming in the stream. The use case for this filter, when input stream contains business name, address, etc., which usually have a small number of tokens. I am guessing, here (test environment) input data containing long paragraphs or documents and that might be causing the issue. Concatenate words from token stream --- Key: SOLR-7193 URL: https://issues.apache.org/jira/browse/SOLR-7193 Project: Solr Issue Type: New Feature Components: Schema and Analysis Reporter: abhishek bafna Attachments: concatenate_words.patch The user entered data often don't have proper spacing between words and words spelling and format also varies from data like business names, address etc. After tokenizing data, we might perform pattern replacement, stop word filtering etc. Later we want to concatenate all the tokens and generate n-grams token for indexing business name and perform the fuzzy match. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-7193) Concatenate words from token stream
[ https://issues.apache.org/jira/browse/SOLR-7193?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14348403#comment-14348403 ] abhishek bafna commented on SOLR-7193: -- The ConcatenateWordsFilter takes all the input token (words) and generate a single token. The CPU time and memory depends on the number and size of the tokens coming in the stream. The use case for this filter, when input stream contains business name, address, etc., which usually have a small number of tokens. I am guessing, here (test environment) input data containing long paragraphs or documents and that might be causing the issue. Concatenate words from token stream --- Key: SOLR-7193 URL: https://issues.apache.org/jira/browse/SOLR-7193 Project: Solr Issue Type: New Feature Components: Schema and Analysis Reporter: abhishek bafna Attachments: concatenate_words.patch The user entered data often don't have proper spacing between words and words spelling and format also varies from data like business names, address etc. After tokenizing data, we might perform pattern replacement, stop word filtering etc. Later we want to concatenate all the tokens and generate n-grams token for indexing business name and perform the fuzzy match. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-7193) Concatenate words from token stream
[ https://issues.apache.org/jira/browse/SOLR-7193?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14348272#comment-14348272 ] chengyunyun commented on SOLR-7193: --- pressure test environment: • Client used Spring restful, • User:500 • Total data:3 million • Problem: solr log search time is only 30ms; but searchClient execute server.query(SolrQuery query) need more time,even 30s;time gap is very large. is it related to CPU or memory? Concatenate words from token stream --- Key: SOLR-7193 URL: https://issues.apache.org/jira/browse/SOLR-7193 Project: Solr Issue Type: New Feature Components: Schema and Analysis Reporter: abhishek bafna Attachments: concatenate_words.patch The user entered data often don't have proper spacing between words and words spelling and format also varies from data like business names, address etc. After tokenizing data, we might perform pattern replacement, stop word filtering etc. Later we want to concatenate all the tokens and generate n-grams token for indexing business name and perform the fuzzy match. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org