different error arisen sometimes

[imElliottt]

I ran the example notebook using one of the CVs provided. I try to build graph for the cv and two job offers.

1. I build graph for the cv succesfully most of the time, but TypeError is arisen sometimes.
   itext2kg/irelations_extraction/irelations_extractor.py, line 96.
   'relationship' is supposed to be a dict, but it's a str sometimes.

2. I tried to build graph for the twe job offers so many times and succeeded only once.
   a. Same problem as above, part of relation extractor
   b. I often fail when try to extracte entities for the second text of the job offer. /itext2kg/itext2kg.py, line 85
   /itext2kg/ientities_extraction/ientities_extractor.py, line 71.
   the para 'name' is supposed to be a str, but it could be a list sometimes.

I think the reason for these problems has something to do with the model I'm using, llama3-8b, because the semantic block, entities and relations I got using this model are not always the same. But I think these problems(unexpected type) should be considered.

Thanks for your patient.

[lairgiyassir]

Hello,

Thank you for pointing out this issue.
Which version of iText2KG are you using ? please check that you are using the latest version.

[imElliottt]

Thanks for your relpy!

itext2kg-0.0.7. I think it's the latest version.
So I built the kg separately for the two offers. job_offer is ok, job_offer_2 didn't work.
Do you have some suggetions for the situation above.

thank you

[Wonder947]

Hi,

I'm meeting the same issue when trying to construct the KG for the second job offer

It gives this error:
"
ValidationError: 1 validation error for Entity
name
Input should be a valid string [type=string_type, input_value=['Translate our core valu...ency and effectiveness'], input_type=list]
"
And it probably originiates from when trying to parse this

"[INFO] ------- Extracting Entities from the Document 2
{'entities': [{'$ref': '#/$defs/Entity', 'label': 'Job Offer', 'name': 'Innovate Design Co. Job Offer'}, {'$ref': '#/$defs/Entity', 'label': 'Company', 'name': 'Innovate Design Co.'}, ..., {'$ref': '#/$defs/Entity', 'label': 'Responsibilities', 'name': ['Translate our core values, product, marketing, and sales objectives into beautifully crafted deliverables', 'Design compelling, brand-aligned digital and print materials, including websites, social media content, ads, third-party marketplaces, presentations, animations, events, prints, etc.', 'Develop and maintain visual brand identity guidelines, ensuring brand consistency across all media and multichannel platforms', "Communicate Innovate Design Co.'s narrative through conversion and data-driven design", 'Participate in brainstorming sessions and collaborate with stakeholders to articulate a creative vision that enhances our brand’s visual storytelling', 'Promote design comprehension and sensibility across the organization, refining work methodologies and design processes to enhance efficiency and effectiveness']}, ....]}"

As it expects the "name" field to be string but get "list" of str.

[JIAQI549]

I had the same problem running "kg_ = itext2kg.build_graph(sections=[semantic_blocks_job_offer, semantic_blocks_job_offer_2], existing_knowledge_graph=kg, ent_threshold=0.6, rel_threshold=0.6)" with llama3：ValidationError: 1 validation error for Entity
name
Input should be a valid string [type=string_type, input_value=['Translate our core valu...ency and effectiveness'], input_type=list]
For further information visit https://errors.pydantic.dev/2.9/v/string_type

[lairgiyassir]

Hello,
Thank you all for your feedback.

The major problem of these local and small models is their inability to structure the output.
The only solution is to upgrade the model size (yes, you will need more resources).

But, we will consider this issue in our next releases ;)