4 ms·

"We were trying to do a boldly cutting edge Natural Language Processing startup, but without a top-notch Natural Language Processing expert." Of the many, many

"We were trying to do a boldly cutting edge Natural Language Processing startup, but without a top-notch Natural Language Processing expert." Of the many, many, many management problems present in this article, this seems like the most atrocious. Attempting to complete a task neither google nor apple nor any big player has tackled is total madness. Where is the phd level research? You have to at least understand the technical aspects of your project at a high level and bring on someone who has a deep grasp of them as a founder. Who wants to build this for you without having the equity and stake? Someone should have just taken a pile of $1.5 million and lit it on fire.

hueving 11y ago
To be fair, if you're in a situation where you want someone 'thinking outside the box', hiring top experts from that field isn't necessarily the right thing. If a person sees all of these horse drawn carriages around and wants to figure out a way to get rid of the horse, he/she doesn't turn to people that are experts at attaching horses to carriages.
- flyinglizard 11y ago
  That's true, and having people with very strong academic bias (e.g. those who spent their short careers obtaining PhD but don't have any industry experience) can be just as bad. You generally want those PhDs as researchers or advisers to your company, not as founders/management. Obviously, exceptions always exist.
Animats 11y ago
"We were trying to do a boldly cutting edge Natural Language Processing startup, but without a top-notch Natural Language Processing expert." It's worse than that. They're trying to do something very hard - a natural language system which can handle a conversation with state, and reliably understands its input. Very, very few systems do that. Things like Siri and Google Voice Search will punt to giving you a web page when they don't understand enough to give a concise answer. Those systems have huge staffs behind them writing rules for special cases. There are commercial products which might make this do-able by a small team. Teneo has a natural language platform good enough for online banking.[1] It can do a conversation with state, and can handle business logic for the bank's services. Their dress ordering demo is amusing.[2] Not only does it do the order, it tries to sell accessories relevant to the dress. That could work; presenting the buyer with a few well-chosen options at the moment of sale is good sales technique. Amazon does that with "Others also bought", but it's not as targeted. But no, they're trying to do it with open source software and nobody with a clue. They also seem to have completely missed that they'll need a back end system which looks at the results from the natural language system, notes when things didn't go well, and can be used to make the system smarter. [1] https://www.youtube.com/watch?v=BwOZaPnJIzQ https://www.youtube.com/watch?v=BwOZaPnJIzQ [2] http://www.artificial-solutions.com/natural-language-interaction-solutions/natural-language-interface/solutions-for-intelligent-websites-applications-and-web-services/ http://www.artificial-solutions.com/natural-language-interac...
- tonylucas 11y ago
  Posting at the direct suggestion of the author of the article (just to clarify I'm not trying to comment hijack), we (www.converse.ai) have solved a good deal of these problems, and are making it available as a platform for companies to build their own conversational services on. We handle: 1) Front end integrations (Slack, Twitter, Intercom, Telegram etc, more to come) 2, Natural language interaction (either via our own NLP system or pluggable into other NLP systems as required) 3, Conversation state and context management 4, Validation and parsing of conversation input based on NLP, Regex, or external functions (e.g calling another API) as required. (NLP isn't best in all scenarios) 5, Static and Dynamic 'Conversation Maps', allowing you to generate them on the fly as required. Comments/Questions etc welcome :)
christoph 11y ago
A standout for me was this nugget early on: Instead of the user clicking a button to edit old Contacts, they would type, “I want to edit the info about my contact Jenny Hei,” and we would then find the info about Jenny Hei and offer it to the user so they could edit it. That was the new plan. This was a brilliant idea. Oh dear...

hueving 11y ago

To be fair, if you're in a situation where you want someone 'thinking outside the box', hiring top experts from that field isn't necessarily the right thing. If a person sees all of these horse drawn carriages around and wants to figure out a way to get rid of the horse, he/she doesn't turn to people that are experts at attaching horses to carriages.

flyinglizard 11y ago
That's true, and having people with very strong academic bias (e.g. those who spent their short careers obtaining PhD but don't have any industry experience) can be just as bad. You generally want those PhDs as researchers or advisers to your company, not as founders/management. Obviously, exceptions always exist.

Animats 11y ago

"We were trying to do a boldly cutting edge Natural Language Processing startup, but without a top-notch Natural Language Processing expert." It's worse than that. They're trying to do something very hard - a natural language system which can handle a conversation with state, and reliably understands its input. Very, very few systems do that. Things like Siri and Google Voice Search will punt to giving you a web page when they don't understand enough to give a concise answer. Those systems have huge staffs behind them writing rules for special cases. There are commercial products which might make this do-able by a small team. Teneo has a natural language platform good enough for online banking.[1] It can do a conversation with state, and can handle business logic for the bank's services. Their dress ordering demo is amusing.[2] Not only does it do the order, it tries to sell accessories relevant to the dress. That could work; presenting the buyer with a few well-chosen options at the moment of sale is good sales technique. Amazon does that with "Others also bought", but it's not as targeted. But no, they're trying to do it with open source software and nobody with a clue. They also seem to have completely missed that they'll need a back end system which looks at the results from the natural language system, notes when things didn't go well, and can be used to make the system smarter. [1] https://www.youtube.com/watch?v=BwOZaPnJIzQ https://www.youtube.com/watch?v=BwOZaPnJIzQ [2] http://www.artificial-solutions.com/natural-language-interaction-solutions/natural-language-interface/solutions-for-intelligent-websites-applications-and-web-services/ http://www.artificial-solutions.com/natural-language-interac...

tonylucas 11y ago
Posting at the direct suggestion of the author of the article (just to clarify I'm not trying to comment hijack), we (www.converse.ai) have solved a good deal of these problems, and are making it available as a platform for companies to build their own conversational services on. We handle: 1) Front end integrations (Slack, Twitter, Intercom, Telegram etc, more to come) 2, Natural language interaction (either via our own NLP system or pluggable into other NLP systems as required) 3, Conversation state and context management 4, Validation and parsing of conversation input based on NLP, Regex, or external functions (e.g calling another API) as required. (NLP isn't best in all scenarios) 5, Static and Dynamic 'Conversation Maps', allowing you to generate them on the fly as required. Comments/Questions etc welcome :)

christoph 11y ago

A standout for me was this nugget early on: Instead of the user clicking a button to edit old Contacts, they would type, “I want to edit the info about my contact Jenny Hei,” and we would then find the info about Jenny Hei and offer it to the user so they could edit it. That was the new plan. This was a brilliant idea. Oh dear...