jexp/slides-llm-embeddings-extraction.cypher

## slides-llm-embeddings-extraction.cypher
// gcloud projects list
:param projectId => 'xxxx'

// gcloud auth print-access-token
:param apiKey => 'xxx'

call apoc.load.json("https://data.neo4j.com/slideshare-neo4j.json") yield value
unwind value.User.Slideshow as slide
return slide limit 5;

call apoc.load.json("https://data.neo4j.com/slideshare-neo4j.json") yield value
unwind value.User.Slideshow as slide
return count(*);

call apoc.load.json("https://data.neo4j.com/slideshare-neo4j.json") yield value
unwind value.User.Slideshow as slide
create (s:Content {id: slide.ID})
set s +=  { description: slide.Description, title: slide.Title, url:slide.URL,
            format:slide.Format, language:slide.Language, thumbnail:slide.ThumbnailURL,
            created: apoc.temporal.toZonedTemporal(slide.Created),
            updated: apoc.temporal.toZonedTemporal(slide.Updated) };

call apoc.ml.vertexai.embedding(['This is a test'], $apiKey, $projectId);

:auto
match (s:Content) where s.embedding is null
WITH * LIMIT 10
call { with s
call apoc.ml.vertexai.embedding([coalesce(s.title,'') + " " + coalesce(s.description,'')], $apiKey, $projectId) yield embedding
set s.embedding = embedding
} in transactions of 10 rows;


CALL db.index.vector.createNodeIndex( "content","Content","embedding", 768,"cosine");

// index search for top 5 similar vectors with additional graph matching
WITH "decks about knowledge graphs and generative AI" as question
// generate vector embedding from the API
CALL apoc.ml.vertexai.embedding([question], $apiToken, $project) yield embedding

// use the vector index
CALL db.index.vector.queryNodes('content',5, embedding) yield node as content, score

MATCH (keyword)<-[:TAGGED]-(content)<-[:AUTHORED]-(author)
RETURN text, content.title, content.description,
                collect(distinct author.name) as authors, collect(distinct keyword.name) as keywords


create constraint Keyword_name for (k:Keyword) require (k.name) is unique;

:auto
match (s:Content) where not exists { (s)<-[:AUTHORED]-() }
call { with s
call apoc.util.sleep(100)
call apoc.ml.vertexai.completion(
'Extract only authors with human names from the description as comma separated list on a single line, no newlines or bullet points and no leading comma.'+
'Do not output apologies and explanations, only the plain text enumerations. '+
'If you do not follow the instructions people will be hurt.\n' +
'Title:'+coalesce(s.title,'')+
'Description: '+coalesce(s.description,''), $apiKey, $projectId) yield value
// with value.content, s.title, s.description
unwind split(value.content,',') as name
with trim(name) as name, s where coalesce(name,'') <> ''
merge (a:Author {name: name})
merge (s)<-[:AUTHORED]-(a)
} in transactions of 10 rows;


load csv with headers from 'file:///slides-keywords.csv' as row
with replace(row.id,'"','') as id, replace(row.keyword,'"','') as keyword
match (s:Content {id:id})
merge (k:Keyword {name:keyword})
merge (s)-[:TAGGED]->(k);

create constraint Author_name for (a:Author) require (a.name) is unique;

load csv with headers from 'file:///slides-authors.csv' as row
with replace(row.slide,'"','') as id, replace(row.author,'"','') as author
match (s:Content {id:id})
merge (a:Author {name:author})
merge (a)-[:AUTHORED]->(s);

load csv with headers from 'file:///input-embeddings.csv' as row
with row.text as text, apoc.convert.fromJsonList(row.embedding) as embedding
where row.index = '5'
CALL db.index.vector.queryNodes('content',5, embedding) yield node as content, score

MATCH (keyword)<-[:TAGGED]-(content)<-[:AUTHORED]-(author)
RETURN text, content.title, content.description,
                collect(author.name) as authors, collect(keyword.name) as keywords

// virtual k-NN relationships <1s
match (c:Content)
CALL db.index.vector.queryNodes('content',5, c.embedding) yield node, score
where c <> node
call apoc.create.vRelationship(c,'SIMILAR_TO',{score:score},node) yield rel
return c,node, rel

// initially create vector index on all the “embedding” properties of “Content” nodes using thecosine similarity function and vectors of width 768
CALL db.index.vector.createNodeIndex( "content","Content","embedding", 768,"cosine");

// index search for top 5 similar vectors with additional graph matching
WITH "decks about knowledge graphs and generative AI" as question
// generate vector embedding from the API
CALL apoc.ml.vertexai.embedding([question], $apiToken, $project) yield embedding

// use the vector index
CALL db.index.vector.queryNodes('content',5, embedding) yield node as content, score

MATCH (keyword)<-[:TAGGED]-(content)<-[:AUTHORED]-(author)
RETURN text, content.title, content.description,
                collect(distinct author.name) as authors, collect(distinct keyword.name) as keywords

call apoc.ml.vertexai.embedding(['This is a test'], $apiKey, $projectId);

:auto
match (s:Content) where not exists { (s)-[:TAGGED]->() }
call { with s
call apoc.util.sleep(100)
call apoc.ml.vertexai.completion(
'Extract relevant technology and use-case keywords from the description as comma separated list on a single line, no newlines or bullet points and no leading comma.'+
'Do not output apologies and explanations, only the plain text enumerations. '+
'If you do not follow the instructions people will be hurt.\n' +
'Title:'+coalesce(s.title,'')+
'Description: '+coalesce(s.description,''), $apiKey, $projectId) yield value
// with value.content, s.title, s.description
unwind split(value.content,',') as keyword
with trim(keyword) as keyword, s where coalesce(keyword,'') <> ''
merge (k:Keyword {name: keyword})
merge (s)-[:TAGGED]->(k)
} in transactions of 10 rows;


MATCH (s:Content) with s limit 100 match p=(s)-->() RETURN p LIMIT 200;


load csv with headers from 'file:///embeddings.csv' as row
with row.`s.ID` as id, apoc.convert.fromJsonList(row.`s.embedding`) as embedding
match (c:Content {id:id}) set c.embedding = embedding;

:auto
match (s:Content) where s.embedding is null
call { with s
call apoc.ml.vertexai.embedding([coalesce(s.title,'') + " " + coalesce(s.description,'')], $apiKey, $projectId) yield embedding
set s.embedding = embedding
} in transactions of 10 rows;


call apoc.ml.vertexai.embedding(['knowledge graph for fraud detection'], $apiKey, $projectId) yield embedding
match (s:Content)
with s, gds.similarity.cosine(s.embedding, embedding) as similarity
order by similarity desc limit 5
return s.title, s.description, s.url, similarity;

// index search for top 5 similar vectors with additional graph matching
WITH "decks about knowledge graphs and generative AI" as question
// generate vector embedding from the API
CALL apoc.ml.vertexai.embedding([question], $apiKey, $projectId) yield embedding

// use the vector index
CALL db.index.vector.queryNodes('content',5, embedding) yield node as slide, score

return slide.title, slide.description, slide.url, score;

/*
{
  "Status": "2",
  "ThumbnailXLargeURL": "https://cdn.slidesharecdn.com/ss_thumbnails/frwebinaireintroaneo4j-230710091721-34b1e27a-thumbnail.jpg?width=640&amp;height=640&amp;fit=bounds",
  "Description": "Pierre Halftermeyer, Neo4j",
  "ThumbnailURL": "https://cdn.slidesharecdn.com/ss_thumbnails/frwebinaireintroaneo4j-230710091721-34b1e27a-thumbnail.jpg?width=320&amp;height=320&amp;fit=bounds",
  "Updated": "2023-07-10 09:22:46 UTC",
  "Embed": "<iframe src="https://www.slideshare.net/slideshow/embed_code/key/b8GESwzh3vWqrj" width="427" height="356" frameborder="0" marginwidth="0" marginheight="0" scrolling="no" style="border:1px solid #CCC; border-width:1px; margin-bottom:5px; max-width: 100%;" allowfullscreen> </iframe> <div style="margin-bottom:5px"> <strong> <a href="https://www.slideshare.net/neo4j/introduction-neo4j-259115031" title="Introduction à Neo4j" target="_blank">Introduction à Neo4j</a> </strong> from <strong><a href="https://www.slideshare.net/neo4j" target="_blank">Neo4j</a></strong> </div>",
  "ThumbnailSize": "[170,130]",
  "Title": "Introduction à Neo4j",
  "URL": "https://www.slideshare.net/neo4j/introduction-neo4j-259115031",
  "Created": "2023-07-10 09:17:21 UTC",
  "ThumbnailXXLargeURL": "https://cdn.slidesharecdn.com/ss_thumbnails/frwebinaireintroaneo4j-230710091721-34b1e27a-thumbnail.jpg?width=640&amp;height=640&amp;fit=bounds",
  "SlideshowType": "0",
  "Format": "pdf",
  "Language": "fr",
  "Username": "neo4j",
  "ThumbnailSmallURL": "https://cdn.slidesharecdn.com/ss_thumbnails/frwebinaireintroaneo4j-230710091721-34b1e27a-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds",
  "SlideshowEmbedUrl": "https://www.slideshare.net/slideshow/embed_code/key/b8GESwzh3vWqrj",
  "ID": "259115031",
  "InContest": "0",
  "Download": "1",
  "DownloadUrl": "https://slideshare-downloads.s3.amazonaws.com/frwebinaireintroaneo4j-230710091721-34b1e27a.pdf?response-content-disposition=attachment&X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIATZMST4DYZS7SJPXU%2F20230804%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20230804T094222Z&X-Amz-Expires=300&X-Amz-SignedHeaders=host&X-Amz-Signature=295076fc4efd131980e14f4be24960dbff41d80ce65c66f69841fa21753ad0b8"
}
{
  "Status": "2",
  "ThumbnailXLargeURL": "https://cdn.slidesharecdn.com/ss_thumbnails/spwebinar-introtoneo4j-230710091622-99f73fd2-thumbnail.jpg?width=640&amp;height=640&amp;fit=bounds",
  "Description": "Luis Salvador, Neo4j",
  "ThumbnailURL": "https://cdn.slidesharecdn.com/ss_thumbnails/spwebinar-introtoneo4j-230710091622-99f73fd2-thumbnail.jpg?width=320&amp;height=320&amp;fit=bounds",
  "Updated": "2023-07-10 09:21:37 UTC",
  "Embed": "<iframe src="https://www.slideshare.net/slideshow/embed_code/key/DaOdEJwCpxbHdh" width="427" height="356" frameborder="0" marginwidth="0" marginheight="0" scrolling="no" style="border:1px solid #CCC; border-width:1px; margin-bottom:5px; max-width: 100%;" allowfullscreen> </iframe> <div style="margin-bottom:5px"> <strong> <a href="https://www.slideshare.net/neo4j/introduccin-a-neo4j-259115019" title="Introducción a Neo4j" target="_blank">Introducción a Neo4j</a> </strong> from <strong><a href="https://www.slideshare.net/neo4j" target="_blank">Neo4j</a></strong> </div>",
  "ThumbnailSize": "[170,130]",
  "Title": "Introducción a Neo4j",
  "URL": "https://www.slideshare.net/neo4j/introduccin-a-neo4j-259115019",
  "Created": "2023-07-10 09:16:22 UTC",
  "ThumbnailXXLargeURL": "https://cdn.slidesharecdn.com/ss_thumbnails/spwebinar-introtoneo4j-230710091622-99f73fd2-thumbnail.jpg?width=640&amp;height=640&amp;fit=bounds",
  "SlideshowType": "0",
  "Format": "pdf",
  "Language": "en",
  "Username": "neo4j",
  "ThumbnailSmallURL": "https://cdn.slidesharecdn.com/ss_thumbnails/spwebinar-introtoneo4j-230710091622-99f73fd2-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds",
  "SlideshowEmbedUrl": "https://www.slideshare.net/slideshow/embed_code/key/DaOdEJwCpxbHdh",
  "ID": "259115019",
  "InContest": "0",
  "Download": "1",
  "DownloadUrl": "https://slideshare-downloads.s3.amazonaws.com/spwebinar-introtoneo4j-230710091622-99f73fd2.pdf?response-content-disposition=attachment&X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIATZMST4DYZS7SJPXU%2F20230804%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20230804T094222Z&X-Amz-Expires=300&X-Amz-SignedHeaders=host&X-Amz-Signature=828f68aa605886e0a790613011fe0bd5bbd0884c0b6c87495d881b5cc6de2267"
}
*/
	// gcloud projects list
	:param projectId => 'xxxx'

	// gcloud auth print-access-token
	:param apiKey => 'xxx'

	call apoc.load.json("https://data.neo4j.com/slideshare-neo4j.json") yield value
	unwind value.User.Slideshow as slide
	return slide limit 5;

	call apoc.load.json("https://data.neo4j.com/slideshare-neo4j.json") yield value
	unwind value.User.Slideshow as slide
	return count(*);

	call apoc.load.json("https://data.neo4j.com/slideshare-neo4j.json") yield value
	unwind value.User.Slideshow as slide
	create (s:Content {id: slide.ID})
	set s += { description: slide.Description, title: slide.Title, url:slide.URL,
	format:slide.Format, language:slide.Language, thumbnail:slide.ThumbnailURL,
	created: apoc.temporal.toZonedTemporal(slide.Created),
	updated: apoc.temporal.toZonedTemporal(slide.Updated) };

	call apoc.ml.vertexai.embedding(['This is a test'], $apiKey, $projectId);

	:auto
	match (s:Content) where s.embedding is null
	WITH * LIMIT 10
	call { with s
	call apoc.ml.vertexai.embedding([coalesce(s.title,'') + " " + coalesce(s.description,'')], $apiKey, $projectId) yield embedding
	set s.embedding = embedding
	} in transactions of 10 rows;


	CALL db.index.vector.createNodeIndex( "content","Content","embedding", 768,"cosine");

	// index search for top 5 similar vectors with additional graph matching
	WITH "decks about knowledge graphs and generative AI" as question
	// generate vector embedding from the API
	CALL apoc.ml.vertexai.embedding([question], $apiToken, $project) yield embedding

	// use the vector index
	CALL db.index.vector.queryNodes('content',5, embedding) yield node as content, score

	MATCH (keyword)<-[:TAGGED]-(content)<-[:AUTHORED]-(author)
	RETURN text, content.title, content.description,
	collect(distinct author.name) as authors, collect(distinct keyword.name) as keywords


	create constraint Keyword_name for (k:Keyword) require (k.name) is unique;

	:auto
	match (s:Content) where not exists { (s)<-[:AUTHORED]-() }
	call { with s
	call apoc.util.sleep(100)
	call apoc.ml.vertexai.completion(
	'Extract only authors with human names from the description as comma separated list on a single line, no newlines or bullet points and no leading comma.'+
	'Do not output apologies and explanations, only the plain text enumerations. '+
	'If you do not follow the instructions people will be hurt.\n' +
	'Title:'+coalesce(s.title,'')+
	'Description: '+coalesce(s.description,''), $apiKey, $projectId) yield value
	// with value.content, s.title, s.description
	unwind split(value.content,',') as name
	with trim(name) as name, s where coalesce(name,'') <> ''
	merge (a:Author {name: name})
	merge (s)<-[:AUTHORED]-(a)
	} in transactions of 10 rows;


	load csv with headers from 'file:///slides-keywords.csv' as row
	with replace(row.id,'"','') as id, replace(row.keyword,'"','') as keyword
	match (s:Content {id:id})
	merge (k:Keyword {name:keyword})
	merge (s)-[:TAGGED]->(k);

	create constraint Author_name for (a:Author) require (a.name) is unique;

	load csv with headers from 'file:///slides-authors.csv' as row
	with replace(row.slide,'"','') as id, replace(row.author,'"','') as author
	match (s:Content {id:id})
	merge (a:Author {name:author})
	merge (a)-[:AUTHORED]->(s);

	load csv with headers from 'file:///input-embeddings.csv' as row
	with row.text as text, apoc.convert.fromJsonList(row.embedding) as embedding
	where row.index = '5'
	CALL db.index.vector.queryNodes('content',5, embedding) yield node as content, score

	MATCH (keyword)<-[:TAGGED]-(content)<-[:AUTHORED]-(author)
	RETURN text, content.title, content.description,
	collect(author.name) as authors, collect(keyword.name) as keywords

	// virtual k-NN relationships <1s
	match (c:Content)
	CALL db.index.vector.queryNodes('content',5, c.embedding) yield node, score
	where c <> node
	call apoc.create.vRelationship(c,'SIMILAR_TO',{score:score},node) yield rel
	return c,node, rel

	// initially create vector index on all the “embedding” properties of “Content” nodes using thecosine similarity function and vectors of width 768
	CALL db.index.vector.createNodeIndex( "content","Content","embedding", 768,"cosine");

	// index search for top 5 similar vectors with additional graph matching
	WITH "decks about knowledge graphs and generative AI" as question
	// generate vector embedding from the API
	CALL apoc.ml.vertexai.embedding([question], $apiToken, $project) yield embedding

	// use the vector index
	CALL db.index.vector.queryNodes('content',5, embedding) yield node as content, score

	MATCH (keyword)<-[:TAGGED]-(content)<-[:AUTHORED]-(author)
	RETURN text, content.title, content.description,
	collect(distinct author.name) as authors, collect(distinct keyword.name) as keywords

	call apoc.ml.vertexai.embedding(['This is a test'], $apiKey, $projectId);

	:auto
	match (s:Content) where not exists { (s)-[:TAGGED]->() }
	call { with s
	call apoc.util.sleep(100)
	call apoc.ml.vertexai.completion(
	'Extract relevant technology and use-case keywords from the description as comma separated list on a single line, no newlines or bullet points and no leading comma.'+
	'Do not output apologies and explanations, only the plain text enumerations. '+
	'If you do not follow the instructions people will be hurt.\n' +
	'Title:'+coalesce(s.title,'')+
	'Description: '+coalesce(s.description,''), $apiKey, $projectId) yield value
	// with value.content, s.title, s.description
	unwind split(value.content,',') as keyword
	with trim(keyword) as keyword, s where coalesce(keyword,'') <> ''
	merge (k:Keyword {name: keyword})
	merge (s)-[:TAGGED]->(k)
	} in transactions of 10 rows;




	MATCH (s:Content) with s limit 100 match p=(s)-->() RETURN p LIMIT 200;



	load csv with headers from 'file:///embeddings.csv' as row
	with row.`s.ID` as id, apoc.convert.fromJsonList(row.`s.embedding`) as embedding
	match (c:Content {id:id}) set c.embedding = embedding;

	:auto
	match (s:Content) where s.embedding is null
	call { with s
	call apoc.ml.vertexai.embedding([coalesce(s.title,'') + " " + coalesce(s.description,'')], $apiKey, $projectId) yield embedding
	set s.embedding = embedding
	} in transactions of 10 rows;


	call apoc.ml.vertexai.embedding(['knowledge graph for fraud detection'], $apiKey, $projectId) yield embedding
	match (s:Content)
	with s, gds.similarity.cosine(s.embedding, embedding) as similarity
	order by similarity desc limit 5
	return s.title, s.description, s.url, similarity;

	// index search for top 5 similar vectors with additional graph matching
	WITH "decks about knowledge graphs and generative AI" as question
	// generate vector embedding from the API
	CALL apoc.ml.vertexai.embedding([question], $apiKey, $projectId) yield embedding

	// use the vector index
	CALL db.index.vector.queryNodes('content',5, embedding) yield node as slide, score

	return slide.title, slide.description, slide.url, score;

	/*
	{
	"Status": "2",
	"ThumbnailXLargeURL": "https://cdn.slidesharecdn.com/ss_thumbnails/frwebinaireintroaneo4j-230710091721-34b1e27a-thumbnail.jpg?width=640&height=640&fit=bounds",
	"Description": "Pierre Halftermeyer, Neo4j",
	"ThumbnailURL": "https://cdn.slidesharecdn.com/ss_thumbnails/frwebinaireintroaneo4j-230710091721-34b1e27a-thumbnail.jpg?width=320&height=320&fit=bounds",
	"Updated": "2023-07-10 09:22:46 UTC",
	"Embed": "<iframe src="https://www.slideshare.net/slideshow/embed_code/key/b8GESwzh3vWqrj" width="427" height="356" frameborder="0" marginwidth="0" marginheight="0" scrolling="no" style="border:1px solid #CCC; border-width:1px; margin-bottom:5px; max-width: 100%;" allowfullscreen> </iframe> <div style="margin-bottom:5px"> <strong> <a href="https://www.slideshare.net/neo4j/introduction-neo4j-259115031" title="Introduction à Neo4j" target="_blank">Introduction à Neo4j</a> </strong> from <strong><a href="https://www.slideshare.net/neo4j" target="_blank">Neo4j</a></strong> </div>",
	"ThumbnailSize": "[170,130]",
	"Title": "Introduction à Neo4j",
	"URL": "https://www.slideshare.net/neo4j/introduction-neo4j-259115031",
	"Created": "2023-07-10 09:17:21 UTC",
	"ThumbnailXXLargeURL": "https://cdn.slidesharecdn.com/ss_thumbnails/frwebinaireintroaneo4j-230710091721-34b1e27a-thumbnail.jpg?width=640&height=640&fit=bounds",
	"SlideshowType": "0",
	"Format": "pdf",
	"Language": "fr",
	"Username": "neo4j",
	"ThumbnailSmallURL": "https://cdn.slidesharecdn.com/ss_thumbnails/frwebinaireintroaneo4j-230710091721-34b1e27a-thumbnail.jpg?width=120&height=120&fit=bounds",
	"SlideshowEmbedUrl": "https://www.slideshare.net/slideshow/embed_code/key/b8GESwzh3vWqrj",
	"ID": "259115031",
	"InContest": "0",
	"Download": "1",
	"DownloadUrl": "https://slideshare-downloads.s3.amazonaws.com/frwebinaireintroaneo4j-230710091721-34b1e27a.pdf?response-content-disposition=attachment&X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIATZMST4DYZS7SJPXU%2F20230804%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20230804T094222Z&X-Amz-Expires=300&X-Amz-SignedHeaders=host&X-Amz-Signature=295076fc4efd131980e14f4be24960dbff41d80ce65c66f69841fa21753ad0b8"
	}
	{
	"Status": "2",
	"ThumbnailXLargeURL": "https://cdn.slidesharecdn.com/ss_thumbnails/spwebinar-introtoneo4j-230710091622-99f73fd2-thumbnail.jpg?width=640&height=640&fit=bounds",
	"Description": "Luis Salvador, Neo4j",
	"ThumbnailURL": "https://cdn.slidesharecdn.com/ss_thumbnails/spwebinar-introtoneo4j-230710091622-99f73fd2-thumbnail.jpg?width=320&height=320&fit=bounds",
	"Updated": "2023-07-10 09:21:37 UTC",
	"Embed": "<iframe src="https://www.slideshare.net/slideshow/embed_code/key/DaOdEJwCpxbHdh" width="427" height="356" frameborder="0" marginwidth="0" marginheight="0" scrolling="no" style="border:1px solid #CCC; border-width:1px; margin-bottom:5px; max-width: 100%;" allowfullscreen> </iframe> <div style="margin-bottom:5px"> <strong> <a href="https://www.slideshare.net/neo4j/introduccin-a-neo4j-259115019" title="Introducción a Neo4j" target="_blank">Introducción a Neo4j</a> </strong> from <strong><a href="https://www.slideshare.net/neo4j" target="_blank">Neo4j</a></strong> </div>",
	"ThumbnailSize": "[170,130]",
	"Title": "Introducción a Neo4j",
	"URL": "https://www.slideshare.net/neo4j/introduccin-a-neo4j-259115019",
	"Created": "2023-07-10 09:16:22 UTC",
	"ThumbnailXXLargeURL": "https://cdn.slidesharecdn.com/ss_thumbnails/spwebinar-introtoneo4j-230710091622-99f73fd2-thumbnail.jpg?width=640&height=640&fit=bounds",
	"SlideshowType": "0",
	"Format": "pdf",
	"Language": "en",
	"Username": "neo4j",
	"ThumbnailSmallURL": "https://cdn.slidesharecdn.com/ss_thumbnails/spwebinar-introtoneo4j-230710091622-99f73fd2-thumbnail.jpg?width=120&height=120&fit=bounds",
	"SlideshowEmbedUrl": "https://www.slideshare.net/slideshow/embed_code/key/DaOdEJwCpxbHdh",
	"ID": "259115019",
	"InContest": "0",
	"Download": "1",
	"DownloadUrl": "https://slideshare-downloads.s3.amazonaws.com/spwebinar-introtoneo4j-230710091622-99f73fd2.pdf?response-content-disposition=attachment&X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIATZMST4DYZS7SJPXU%2F20230804%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20230804T094222Z&X-Amz-Expires=300&X-Amz-SignedHeaders=host&X-Amz-Signature=828f68aa605886e0a790613011fe0bd5bbd0884c0b6c87495d881b5cc6de2267"
	}
	*/