Extracting Emerging Knowledge from Social Media

Today I presented our full paper titled “Extracting Emerging Knowledge from Social Media” at the WWW 2017 conference.

The work is based on a rather obvious assumption, i.e., that knowledge in the world continuously evolves, and ontologies are largely incomplete for what concerns low-frequency data, belonging to the so-called long tail.

Socially produced content is an excellent source for discovering emerging knowledge: it is huge, and immediately reflects the relevant changes which hide emerging entities.

In the paper we propose a method and a tool for discovering emerging entities by extracting them from social media.

Once instrumented by experts through very simple initialization, the method is capable of finding emerging entities; we propose a mixed syntactic + semantic method. The method uses seeds, i.e. prototypes of emerging entities provided by experts, for generating candidates; then, it associates candidates to feature vectors, built by using terms occurring in their social content, and then ranks the candidates by using their distance from the centroid of seeds, returning the top candidates as result.

The method can be continuously or periodically iterated, using the results as new seeds.

The PDF of the full paper presented at WWW 2017 is available online (open access with Creative Common license).

You can also check out the slides of my presentation on Slideshare.

A demo version of the tool is available online for free use, thanks also to our partners Dandelion and Microsoft Azure.

You can TRY THE TOOL NOW if you want.

Keynote from Google Research on Building Knowlege Bases at #ICWE2016

I report here some highlights of the keynote speech by Xin Luna Dong at the 16th International Conference on Web Engineering (ICWE 2016). Incidentally, she is now moving to Amazon for starting a new project on building an Amazon knowledge base.
Building knowledge bases still remains a challenging task.
First, one has to decide how to build the knowledge: automatically or manually?
A survey in 2014 reported the following list of large efforts in knowledge building: the top 4 approaches are manually curated, the bottom 3 are automatic.
Google’s knowledge vault and knowledge Graph are the big winners in terms of volume.
When you move to long tail content, curation does not scale. Automation must be viable and precise.
This is in line with our own research line we are starting on Extracting Changing Knowledge (we presented a short paper at a Web Science 2016 workshop last month). Here is a summary of our approach:
Where knowledge can be extracted from? In Knowledge Valut:
  • largest share of the content comes from DOM structured documents
  • then textual content
  • then annotated content
  • and a small share from web tables

Knowledge Vault is a matrix based approach to knowledge building, with rows = entities and columns= attributes.

It assumes the entities to be available (e.g. in Freebase), and builds a training over that.
One can build KBs by building buckets of triples, with similar probability of being correct. It’s important to precisely estimate correctness probability.
Errors can include mistakes on:
  • triple identification
  • entity linkage
  • predicate linkage
  • source data

Besides general purpose KBs, Google built lightweight vertical knowledge bases (more than 100 available now).

When extracting knowledge, the ingredients are: datasource, extractor approach, the data items themselves, facts and their probability of truth.

Several models can be used for extracting knowledge. Two extremes of the spectrum are:

  1. Single-truth model. Every fact has only one truth. We trust the value of the highest number of datasources.
  2. Multilaeyer model. separates source quality from extractor quality and data errors from extraction errors. One can build a knowledge-based trust model, defining trustworthiness of web pages. One can compare this measure with respect to page rank of web pages:

In general, the challenge is to move from individual information and data points, to integrated and connected knowledge. Building the right edges is really hard though.
Overall, a lot of ingredients influence the correctness of knowledge: temporal aspects, data source correctness, capability of extraction and validation, and so on–

In summary: Plenty of research challenges to be addressed, both by the datascience and modeling communities!

To keep updated on my activities you can subscribe to the RSS feed of my blog or follow my twitter account (@MarcoBrambi).

A bottom-up, knowledge-aware approach to integrating and querying web data services – ACM Trans. on the Web

The October 2013 issue of the ACM Transaction on the Web includes an article of ours on bottom-up domain model design of connected web data sources. This is becoming a more and more important problem as a wealth of data services is becoming available on the Web. Indeed, building and querying Web applications that effectively integrate Web content is increasingly important. However, schema integration and ontology matching with the aim of registering data services often requires a knowledge-intensive, tedious, and error-prone manual process. In the paper we tackle this issue as described below.

The paper has been authored by Stefano Ceri, Silvia Quarteroni and myself within the research project Search Computing.

The full paper is available for download on the ACM Digital Library (free of charge, courtesy of the ACM Author-izer service) through this URL:


This is the summary of the contribution:

We present a bottom-up, semi-automatic service registration process that refers to an external knowledge base and uses simple text processing techniques in order to minimize and possibly avoid the contribution of domain experts in the annotation of data services. The first by-product of this process is a representation of the domain of data services as an entity-relationship diagram, whose entities are named after concepts of the external knowledge base matching service terminology rather than being manually created to accommodate an application-specific ontology. Second, a three-layer annotation of service semantics (service interfaces, access patterns, service marts) describing how services “play” with such domain elements is also automatically constructed at registration time. When evaluated against heterogeneous existing data services and with a synthetic service dataset constructed using Google Fusion Tables, the approach yields good results in terms of data representation accuracy.

We subsequently demonstrate that natural language processing methods can be used to decompose and match simple queries to the data services represented in three layers according to the preceding methodology with satisfactory results. We show how semantic annotations are used at query time to convert the user’s request into an executable logical query. Globally, our findings show that the proposed registration method is effective in creating a uniform semantic representation of data services, suitable for building Web applications and answering search queries.

The bibtex reference is as follows:

author = {Quarteroni, Silvia and Brambilla, Marco and Ceri, Stefano},
title = {A bottom-up, knowledge-aware approach to integrating and querying web data services},
journal = {ACM Trans. Web},
issue_date = {October 2013},
volume = {7},
number = {4},
month = nov,
year = {2013},
issn = {1559-1131},
pages = {19:1--19:33},
articleno = {19},
numpages = {33},
url = {http://doi.acm.org/10.1145/2493536},
doi = {10.1145/2493536},
acmid = {2493536},
publisher = {ACM},
address = {New York, NY, USA},
keywords = {Web data integration, Web data services, Web services, natural language Web query, service querying, structured Web search},

To keep updated on my activities you can subscribe to the RSS feed of my blog or follow my twitter account (@MarcoBrambi).