SPARC has released an excellent primer on Article-Level Metrics.
Important characteristics of ALM cited in the primer:
Traditionally, scholarly article influence was measured by the impact of the journal which originally published the work. For a description of that process, see our guide, Assessing Journal Quality.
There are many reasons to measure impact of individual authors or articles apart from the journals in which they publish. Greater speed of feedback and superior relationship mapping and influence tracking can be accomplished at the article level.
"The prevalence for online journal usage data means that alternative metrics to traditional citations now play a significant role in assessing (with greater immediacy) the use, impact and wider reach of a research article. Making these metrics available at the article level also recognizes the need for research to be judged independently of the journal in which it has been published." IOPscience
PLoS Article-Level Metrics measure the dissemination and reach of published research articles.
Traditionally, the impact of research articles has been measured by the publication journal. But a more informative view is one that examines the overall performance and reach of the articles themselves. Article-Level Metrics are a comprehensive set of impact indicators that enable numerous ways to assess and navigate research most relevant to the field itself, including:
Article-Level Metrics are available, upon publication, for every article published by PLOS.
"The journal is built around the delivery of ink and paper by horses and boats. Today, we have better ink and faster horses, but no fundamental change. This change, especially in an institution as conservative as the academy, is not easy and takes time.
We suggest that this revolution will result in a more diverse and decentralized metajournal. In this DcJ, authors will publish any sort of product they create. They will adapt their work's form and make it retrievable with the help of external service providers. They will market it over richly connected networks with the help of specialists or without. They will certify it in dozens of ways, using hundreds or thousands of competing stamping and ranking agencies and algorithms. And all this data will be managed, organized, and curated by a set of relevance and ranking tools that will present customized views of the metajournal for scholars, practitioners, and administrators alike."
Priem J and Hemminger BM (2012) Decoupling the scholarly journal. Front. Comput. Neurosci. 6:19. doi: 10.3389/fncom.2012.00019